1 00:00:00,212 --> 00:00:03,370 Hey, Frank here. I just want to let everyone know real quick that I will 2 00:00:03,402 --> 00:00:07,093 be going to KubeCon this week. So if you're there, be sure to say 3 00:00:07,093 --> 00:00:10,591 hello. And if you're not there, you can live vicariously through 4 00:00:10,703 --> 00:00:14,507 me and check out the coverage that I'm going 5 00:00:14,507 --> 00:00:17,058 to do. All right. And with that, let's get on to the show. 6 00:00:28,330 --> 00:00:31,432 Well, hello and welcome back to Data Driven, the podcast where we explore the emerging 7 00:00:31,560 --> 00:00:35,119 industry of artificial intelligence, data science, and of course, 8 00:00:35,311 --> 00:00:38,742 none of this is all possible without data engineers. 9 00:00:39,399 --> 00:00:43,022 Unfortunately, my favorite data engineer in the world could not make it 10 00:00:43,022 --> 00:00:46,388 today, so I'm going solo today. And 11 00:00:46,773 --> 00:00:49,274 I am speaking today with Praparna 12 00:00:49,995 --> 00:00:53,810 Moharana, and hopefully I said that right, who is a reporting 13 00:00:53,906 --> 00:00:57,689 analyst at Tempus Unlimited, and she develops healthcare analytics 14 00:00:57,689 --> 00:00:59,650 and compliance reporting solutions She's a senior consultant at Accenture, where she develops and implements 15 00:00:59,650 --> 00:01:02,780 data solutions that transform complex operational data into 16 00:01:02,780 --> 00:01:06,000 actionable business insights. She specializes in 17 00:01:06,000 --> 00:01:09,580 SQL, Power BI, data modeling, ETL, and 18 00:01:09,580 --> 00:01:13,420 healthcare reporting with a focus on improving transparency, 19 00:01:13,480 --> 00:01:17,080 compliance monitoring, and decision-making within Medicaid and healthcare 20 00:01:17,080 --> 00:01:20,600 programs. And just because she's here doesn't mean any 21 00:01:20,600 --> 00:01:24,300 statements she makes is a— Her 22 00:01:24,300 --> 00:01:28,114 statements are not matching, may not reflect the opinions of 23 00:01:28,162 --> 00:01:30,385 her employer. Is that a good safety, uh, 24 00:01:32,043 --> 00:01:35,425 thing there? Yes. Awesome, awesome. 25 00:01:36,552 --> 00:01:40,150 So very interesting. First off, tell me 26 00:01:40,166 --> 00:01:43,974 what EVV is, because that's a new term I had not heard 27 00:01:44,038 --> 00:01:47,810 of. And, um, for those who are longtime fans of the 28 00:01:47,826 --> 00:01:51,480 show, and dare I say readers of my blog, I have a 29 00:01:51,480 --> 00:01:55,263 special place in my heart for electronic medical records. That's how I got into the 30 00:01:55,263 --> 00:01:58,454 tablet PC. platform. So what is EVV? 31 00:01:59,995 --> 00:02:03,626 So EVV stands for Electronic Visit Verification. 32 00:02:04,324 --> 00:02:07,471 So it is a technology which is used to 33 00:02:08,004 --> 00:02:11,312 track home-based care services. For 34 00:02:11,312 --> 00:02:14,488 example, if a caregiver is scheduled to visit a 35 00:02:14,616 --> 00:02:18,298 patient, so it will verify the clock-in and 36 00:02:18,298 --> 00:02:21,931 clock-out time of the caregiver and track 37 00:02:22,436 --> 00:02:24,904 Using the technology called EVV. 38 00:02:26,228 --> 00:02:30,023 Okay, interesting. So it's kind of like home healthcare visits and things like that? 39 00:02:31,058 --> 00:02:33,468 Yes. Okay. What— 40 00:02:35,054 --> 00:02:38,499 I'm unfamiliar with this, and I wonder how many of our listeners are too. What 41 00:02:38,499 --> 00:02:42,265 do you capture? And what is— why does 42 00:02:42,314 --> 00:02:46,010 getting it right matter? So mostly each 43 00:02:46,059 --> 00:02:49,629 of the patients would be provided with different kind of 44 00:02:49,678 --> 00:02:53,396 caregivers. And each day, each of the patients— and 45 00:02:53,494 --> 00:02:57,153 every day, the EVV system will be tracking the 46 00:02:57,153 --> 00:03:00,567 clock-in, clock-out of all the caregivers, and we 47 00:03:00,696 --> 00:03:04,222 collect the timesheet data. So based on that, the 48 00:03:04,222 --> 00:03:08,003 caregivers are being paid. So we are collecting 49 00:03:08,343 --> 00:03:11,803 all the timesheet data basically using all the EVV 50 00:03:12,158 --> 00:03:15,771 systems. So we need to make sure that the 51 00:03:15,987 --> 00:03:19,617 .evv systems are documenting correct reports. 52 00:03:20,424 --> 00:03:23,780 There should be no negative discrepancy during 53 00:03:23,974 --> 00:03:27,629 duration, or that we are getting the 54 00:03:27,694 --> 00:03:30,791 capture duration in a proper document manner. 55 00:03:31,423 --> 00:03:34,342 And you have to follow the sort of healthcare compliance 56 00:03:34,342 --> 00:03:37,651 requirements based on the particular 57 00:03:37,732 --> 00:03:41,463 organization. So this way, uh, we analyze the 58 00:03:41,545 --> 00:03:44,874 EDV data And document it properly and do the 59 00:03:44,874 --> 00:03:48,676 validation. Interesting. How did 60 00:03:48,676 --> 00:03:51,879 you find your way into data analytics and data engineering? 61 00:03:53,760 --> 00:03:57,480 So it is a bit of both, I would say. It's a long 62 00:03:57,917 --> 00:04:01,686 journey. I didn't plan from the beginning or think about it, 63 00:04:02,430 --> 00:04:06,146 and I'm really grateful that I came across data in 64 00:04:06,243 --> 00:04:10,040 my journey while exploring my career options. So basically it began 65 00:04:10,040 --> 00:04:13,760 from 11th and 12th grade when I had elected 66 00:04:13,760 --> 00:04:17,540 for computer science. And I have other subjects also, but 67 00:04:17,540 --> 00:04:21,320 I was so, uh, happy that like you study a small 68 00:04:21,320 --> 00:04:24,919 concept in computer science, you are done, and you just apply 69 00:04:24,919 --> 00:04:28,240 across various problem statements. It's so easy. So 70 00:04:28,240 --> 00:04:31,360 that's why I did my undergrad in computer 71 00:04:31,360 --> 00:04:35,100 science. But in computer science, I was a little 72 00:04:35,100 --> 00:04:37,905 bit overwhelmed because of exploring different 73 00:04:38,163 --> 00:04:41,966 areas of computer science and different 74 00:04:42,030 --> 00:04:45,673 programming languages and web applications, I would 75 00:04:45,755 --> 00:04:49,451 overrun. But later I was quite determined I 76 00:04:49,565 --> 00:04:53,258 will go for a higher level degree. And during my gap 77 00:04:53,338 --> 00:04:57,117 year, COVID happened, and during my gap year I took data science 78 00:04:57,117 --> 00:05:00,088 course and I learned 79 00:05:00,201 --> 00:05:04,045 about exploratory data analysis, data quality, 80 00:05:04,110 --> 00:05:07,580 data cleaning. Machine learning methods I 81 00:05:07,580 --> 00:05:11,160 learned. I, and I was really interested in, 82 00:05:11,160 --> 00:05:14,800 uh, that, and it, and it, uh, applies, um, naturally 83 00:05:14,800 --> 00:05:18,140 with my instincts because I am good in observing and noticing 84 00:05:18,140 --> 00:05:21,539 things. So I was really find it very 85 00:05:21,539 --> 00:05:25,340 fascinating. So instead, when I saw my curriculum 86 00:05:25,340 --> 00:05:28,640 for master's in computer science, I didn't saw any data 87 00:05:28,640 --> 00:05:32,220 analysis or data science course. I thought it's better 88 00:05:32,220 --> 00:05:35,735 to switch for master's in data analytics previously, 89 00:05:35,911 --> 00:05:39,534 and that way data found me and I found 90 00:05:39,758 --> 00:05:42,884 data. It's a mutual relationship and it started. 91 00:05:43,541 --> 00:05:46,651 That's cool. Like, that's the question I always like to ask is like, did you 92 00:05:46,651 --> 00:05:50,386 find data or did data find you? And it sounds like 93 00:05:50,450 --> 00:05:53,912 it was a match made in heaven, I suppose, because you— yeah, no. 94 00:05:54,281 --> 00:05:58,112 And data can be very— from the first glance of it, for me, it wasn't 95 00:05:58,128 --> 00:06:01,970 love at first sight. It was kind of like, It 96 00:06:01,970 --> 00:06:04,884 was kind of like, I kind of saw it again in a fresh light. 97 00:06:06,002 --> 00:06:09,030 And I realized like, there's a really good 98 00:06:09,645 --> 00:06:13,270 engineering discipline behind this. And there's so many 99 00:06:13,303 --> 00:06:17,025 problems you can solve with advanced analytics and things like that. 100 00:06:19,703 --> 00:06:23,470 So my notes say that you do independent research. What, what questions are 101 00:06:23,518 --> 00:06:27,232 you— what are you exploring in your independent research? So mostly in 102 00:06:27,424 --> 00:06:31,094 my independent research, I mostly deal 103 00:06:31,174 --> 00:06:34,876 with collaborating with other industrial people. For 104 00:06:34,876 --> 00:06:38,545 example, I collaborated in how to build 105 00:06:38,545 --> 00:06:42,006 and like whether it is beneficial to 106 00:06:42,631 --> 00:06:46,157 go for to discuss in our research and help people 107 00:06:46,157 --> 00:06:49,906 across industries in topics like it is, is 108 00:06:50,019 --> 00:06:53,867 it better to build an in-house BI tools in place 109 00:06:53,867 --> 00:06:56,713 of vendor-based. We discussed that in our 110 00:06:57,037 --> 00:07:00,530 collaboration and when to go for 111 00:07:00,676 --> 00:07:03,992 in-house BI development and what kind of 112 00:07:04,057 --> 00:07:07,227 customization you can have with in-house BI 113 00:07:07,308 --> 00:07:10,897 developments and what kind of— for example, I've got 114 00:07:10,897 --> 00:07:14,535 someone who is, who has knowledge of Salesforce, 115 00:07:14,551 --> 00:07:18,077 somehow who has knowledge of AI intelligence, and we would 116 00:07:18,109 --> 00:07:21,409 collaborate that how with integration of 117 00:07:21,409 --> 00:07:25,034 Salesforce and the data architecture knowledge I 118 00:07:25,114 --> 00:07:28,868 have, we can collaborate and make the architecture and 119 00:07:28,965 --> 00:07:32,283 people can use that architecture and 120 00:07:32,283 --> 00:07:36,133 Salesforce collaboration across industries and 121 00:07:36,214 --> 00:07:39,926 AI intelligence tools. So this way we try 122 00:07:39,926 --> 00:07:43,059 to help, like even in my speaking opportunities, I 123 00:07:43,139 --> 00:07:46,498 extensively talked about how we can 124 00:07:46,620 --> 00:07:49,988 automate like all the reports using tools like 125 00:07:49,988 --> 00:07:52,005 SQL, Python, and Power BI, 126 00:07:53,913 --> 00:07:57,175 and why it is taking exact various 127 00:07:57,207 --> 00:08:00,356 examples. For example, healthcare compliance, for 128 00:08:00,356 --> 00:08:04,199 example, and how to handle voluminous 129 00:08:04,199 --> 00:08:07,869 amount of data, why it is really important to have an end-to-end 130 00:08:08,126 --> 00:08:11,764 architecture, and based on your data source, what 131 00:08:11,764 --> 00:08:14,876 kind of tools you have to decide. For 132 00:08:14,876 --> 00:08:18,480 example, if you are collecting data 133 00:08:18,480 --> 00:08:22,218 from API calls, why it's better to choose 134 00:08:22,395 --> 00:08:26,221 Python if you have diverse data source, because it 135 00:08:26,253 --> 00:08:29,838 would provide a large source of diversity and flexibility 136 00:08:30,338 --> 00:08:33,888 while doing data monitoring or data log, like 137 00:08:33,888 --> 00:08:37,519 detecting the errors in the pipeline. It will be 138 00:08:37,600 --> 00:08:40,910 easier, and even API validation, authentication, 139 00:08:40,910 --> 00:08:44,690 tokenization. If you choose Python and if you have 2 140 00:08:44,690 --> 00:08:48,350 databases, it's best to go with linked servers and 141 00:08:48,350 --> 00:08:52,150 open queries because if you use Python for that, what happens 142 00:08:52,150 --> 00:08:55,770 if you have 2 open cursors from different databases? 143 00:08:56,090 --> 00:08:59,070 If cursor 1 is running a select statement, 144 00:08:59,870 --> 00:09:03,430 so what happens? The cursor 2 open, so 145 00:09:03,430 --> 00:09:07,250 it— meanwhile the cursor 1 is executing the 146 00:09:07,250 --> 00:09:10,770 query, the cursor 2 is dead. And because of that, you 147 00:09:10,770 --> 00:09:13,730 would have data latency issues 148 00:09:13,730 --> 00:09:17,450 and you won't get timely reports. So it's better to 149 00:09:17,450 --> 00:09:20,730 go with SQL Server. So you have to be smart around 150 00:09:20,730 --> 00:09:24,550 deciding the tools, whether you should go with Python or 151 00:09:24,550 --> 00:09:28,270 SQL. And because mostly I talk about data 152 00:09:28,270 --> 00:09:31,470 automation, means automation of 153 00:09:31,470 --> 00:09:35,220 reports which are manually generated into Excel, 154 00:09:35,500 --> 00:09:39,165 So whenever people think about reports, they mostly 155 00:09:39,615 --> 00:09:43,441 go around a basic idea or think about in generalized way 156 00:09:43,569 --> 00:09:47,329 about dashboards, summary reports, charts, but it is 157 00:09:47,361 --> 00:09:51,042 not about that. It's entire lifecycle of data from 158 00:09:51,317 --> 00:09:55,133 data extraction, making of reports, applying, knowing the 159 00:09:55,133 --> 00:09:57,906 business rules the stakeholders want, 160 00:09:59,781 --> 00:10:03,240 and providing reports to using Power BI, and 161 00:10:03,240 --> 00:10:06,900 even you can even automate reports further using 162 00:10:06,900 --> 00:10:10,540 Power BI if the stakeholders need email 163 00:10:10,540 --> 00:10:13,820 like more than 8 refreshes in Power BI 164 00:10:13,820 --> 00:10:17,520 overnight. Like Power BI can do 8 refreshes, but if 165 00:10:17,520 --> 00:10:20,840 you want more than 8 refreshes, you should use Power 166 00:10:20,840 --> 00:10:24,640 Automate or various things. If you want to split the file 167 00:10:24,640 --> 00:10:28,340 into various categories and you can't provide 168 00:10:28,340 --> 00:10:31,676 like Power BI Pro licenses, licenses to 169 00:10:31,676 --> 00:10:35,204 various users, you can sure put that in a 170 00:10:35,204 --> 00:10:38,893 SharePoint, pen drive, or OneDrive, whatever you use, and 171 00:10:39,330 --> 00:10:43,062 the Power Automate can split it. And even you can email it. 172 00:10:43,127 --> 00:10:46,906 So Power Automate can further automate the work which we 173 00:10:46,906 --> 00:10:49,953 manually do in Excel, like splitting of files 174 00:10:50,934 --> 00:10:54,003 or like emailing the report. So 175 00:10:54,582 --> 00:10:58,230 mostly I am more interested to the automation parts of 176 00:10:58,230 --> 00:11:01,165 reporting and Making things easier. 177 00:11:02,367 --> 00:11:05,958 Interesting. And when you say cursor, you don't— because we live in the AI 178 00:11:06,038 --> 00:11:09,420 age that's dominated by AI tools, you don't mean cursor the 179 00:11:09,901 --> 00:11:12,899 AI tool, you mean the cursor of the old-school SQL 180 00:11:13,748 --> 00:11:17,563 process, right? Mostly it is 181 00:11:17,595 --> 00:11:21,138 being processed sequentially, so it's better to go 182 00:11:21,282 --> 00:11:24,777 with— but I won't say because of AI, because many 183 00:11:25,033 --> 00:11:28,704 organizations still refrain from using AI because of 184 00:11:28,720 --> 00:11:32,430 high sensitivity and confidential data. refrain from using 185 00:11:32,430 --> 00:11:36,170 AI tools and really against Copilots, and, 186 00:11:36,170 --> 00:11:39,850 uh, they totally don't use that. So the thing is, I was 187 00:11:39,850 --> 00:11:43,630 talking about cursors because in Python we sequentially 188 00:11:43,630 --> 00:11:47,330 would be running a line of code. So it's better if 189 00:11:47,330 --> 00:11:51,170 we use open query, it won't run sequentially. Like both, we would have, 190 00:11:51,170 --> 00:11:54,850 uh, both the active connection. In cursors, one cursor 191 00:11:54,850 --> 00:11:58,250 is running, so other cursor would be dead when 192 00:11:58,250 --> 00:12:01,935 first cursor is running. That That's the issue I 193 00:12:02,435 --> 00:12:05,369 had faced while troubleshooting a lot, which I 194 00:12:05,901 --> 00:12:09,077 discovered from my own when I used 2. Yeah, no one— 195 00:12:09,528 --> 00:12:13,299 every time I mentioned doing a— using a 196 00:12:13,299 --> 00:12:16,955 cursor, they always— everyone's told me do 197 00:12:16,955 --> 00:12:20,095 something else, right? It's really, you know, 198 00:12:20,948 --> 00:12:24,780 I think I've only ever used it professionally twice in my 199 00:12:24,828 --> 00:12:28,431 career. And once I got yelled at for 200 00:12:28,431 --> 00:12:32,076 doing it, and then the other time was someone who was far 201 00:12:32,140 --> 00:12:35,801 more experienced than me said, no, that really is the only use 202 00:12:35,801 --> 00:12:39,510 case. Like, that really was the only solution. So I think that's— it's 203 00:12:39,526 --> 00:12:42,930 very telling kind of how that worked out. But 204 00:12:44,086 --> 00:12:47,683 in practical terms, what does good data 205 00:12:47,683 --> 00:12:48,598 quality look like? 206 00:12:51,569 --> 00:12:55,266 So mostly good data quality will look like 207 00:12:55,555 --> 00:12:59,187 that each— if you look at a single 208 00:12:59,187 --> 00:13:02,493 record in your database, it would have all the 209 00:13:02,719 --> 00:13:05,771 accurate, no missing values in a single record, 210 00:13:06,352 --> 00:13:09,565 and there wouldn't be any kind of 211 00:13:09,823 --> 00:13:13,357 data discrepancy like negative, for example, 212 00:13:13,389 --> 00:13:17,181 duration is negative 15 minutes, you need to validate the 213 00:13:17,181 --> 00:13:20,718 data. So I would I would say missing— no 214 00:13:20,718 --> 00:13:24,376 missing values. All the fields should have values in structured data. 215 00:13:25,762 --> 00:13:29,108 And I would say that's the most important part of data 216 00:13:29,173 --> 00:13:32,923 quality. And when you go through the 217 00:13:32,972 --> 00:13:36,757 entire lifecycle of data, there should be no duplicate records. 218 00:13:38,222 --> 00:13:41,382 You should have each unique record. 219 00:13:41,768 --> 00:13:45,444 That's the most important thing about data quality, 220 00:13:45,556 --> 00:13:49,057 I would say. And like while building architecture, 221 00:13:49,573 --> 00:13:53,052 like the entire lifecycle I was talking about, when you pull data from the raw 222 00:13:53,197 --> 00:13:56,777 source, you're moving from SQL layer to Python 223 00:13:56,890 --> 00:14:00,407 layer to the reporting layer like Power BI. You have to 224 00:14:00,472 --> 00:14:04,086 see that the data which you have collected, if 225 00:14:04,150 --> 00:14:07,079 it is 1,000 rows, it should be 226 00:14:07,079 --> 00:14:10,876 1,000 rows till the end of the reporting 227 00:14:10,908 --> 00:14:14,359 when you reporting layer, taking into 228 00:14:14,456 --> 00:14:17,647 consideration how you validate the record. 229 00:14:19,630 --> 00:14:23,455 Why do teams struggle so often with data quality? Because data quality 230 00:14:23,616 --> 00:14:27,296 has been a problem that since I 231 00:14:27,393 --> 00:14:31,104 was a young man, I had way less gray hair, had been an issue. It's 232 00:14:31,121 --> 00:14:33,896 like, so why are we struggling with it still? 233 00:14:35,541 --> 00:14:38,349 I remember when the discussion was, 234 00:14:40,234 --> 00:14:44,063 Does data quality matter? Clearly it does, and I think 235 00:14:44,063 --> 00:14:47,575 we found out the hard way that it does. But why is this so hard 236 00:14:47,575 --> 00:14:51,183 to not have missing values and things like that? What 237 00:14:51,600 --> 00:14:55,369 is it, industry use case specific? What's your thoughts on that? 238 00:14:56,395 --> 00:15:00,180 For me, it depends how we— how the source is collecting the 239 00:15:00,180 --> 00:15:03,114 data, right? Uh, maybe we could have 240 00:15:03,243 --> 00:15:06,898 inaccurate time zones. Because the, for 241 00:15:06,898 --> 00:15:10,536 example, system would be in a different time zone in UTC and we 242 00:15:10,584 --> 00:15:14,301 are, we want in EST time zone where we are. 243 00:15:14,910 --> 00:15:18,019 So mostly time zone issues and 244 00:15:18,933 --> 00:15:22,522 it could be like we would have negative, 245 00:15:22,602 --> 00:15:26,016 like missing values as I mentioned earlier, and 246 00:15:26,448 --> 00:15:30,294 we need values in standardized date format. We 247 00:15:30,310 --> 00:15:34,140 would always face issues because we are trying to collect data from 248 00:15:34,140 --> 00:15:37,520 different sources, not from a single 249 00:15:37,520 --> 00:15:41,060 centralized unit. And we are trying to 250 00:15:41,060 --> 00:15:44,060 gather information to get an 251 00:15:44,060 --> 00:15:47,580 answer to a single question which we 252 00:15:47,580 --> 00:15:49,700 want, as the stakeholder 253 00:15:49,700 --> 00:15:53,080 wants. I think mostly because we are 254 00:15:53,080 --> 00:15:56,780 trying to collect it from different systems. From one would be coming from 255 00:15:56,780 --> 00:16:00,100 API call, one would be coming from a different 256 00:16:00,100 --> 00:16:03,794 database system. So mostly because of different systems, we 257 00:16:03,794 --> 00:16:07,527 would always face data quality issues. You 258 00:16:07,559 --> 00:16:10,731 are on mute. I'm sorry. 259 00:16:11,548 --> 00:16:13,855 Do you think AI can help with data quality issues? 260 00:16:15,409 --> 00:16:19,222 Yes, they can help with data quality issues. 261 00:16:19,398 --> 00:16:23,099 But as far as I'm aware, I'm reading through 262 00:16:23,403 --> 00:16:27,088 other people's experience, and from what 263 00:16:27,088 --> 00:16:30,885 I've read about AI is that when you try to 264 00:16:30,933 --> 00:16:34,362 solve something complex, problem, it's really 265 00:16:34,795 --> 00:16:38,081 difficult for AI to keep track of the data 266 00:16:38,081 --> 00:16:41,382 quality and what we want. So 267 00:16:41,847 --> 00:16:45,582 we really need the human judgment to ensure that 268 00:16:45,598 --> 00:16:49,204 what kind of standardization and normalization we need to 269 00:16:49,284 --> 00:16:52,666 apply to our data for improving the data 270 00:16:52,666 --> 00:16:56,352 quality. We can take help of AI to 271 00:16:56,352 --> 00:16:59,654 improve the process or take suggestion, but it 272 00:16:59,782 --> 00:17:03,463 can't hopefully help with automating the data quality, 273 00:17:03,576 --> 00:17:06,525 like, by any human 274 00:17:06,735 --> 00:17:10,320 interaction or human reviewing it manually. 275 00:17:12,151 --> 00:17:12,609 Interesting. 276 00:17:16,185 --> 00:17:19,998 What's your take on the use of AI in 277 00:17:20,110 --> 00:17:23,153 data systems, like, in the data engineering space? Is that— 278 00:17:25,861 --> 00:17:29,270 I think data engineers have a pretty bright 279 00:17:29,286 --> 00:17:32,881 future in terms of career-wise, but 280 00:17:33,558 --> 00:17:36,073 how do you think the rise of AI agents are gonna 281 00:17:38,020 --> 00:17:39,286 impact people's careers? 282 00:17:41,655 --> 00:17:45,398 Um, it would— it wouldn't— 283 00:17:45,543 --> 00:17:49,023 I think from my point of view, it can't replace— 284 00:17:49,605 --> 00:17:52,924 like, it is mostly said that AI will replace 285 00:17:53,192 --> 00:17:56,679 human jobs. But I feel it 286 00:17:56,872 --> 00:18:00,504 won't actually completely replace. It could 287 00:18:01,050 --> 00:18:03,942 make data engineers' life easier 288 00:18:04,890 --> 00:18:08,730 because in certain directions 289 00:18:08,843 --> 00:18:12,523 we could use agentic AIs and AI 290 00:18:13,101 --> 00:18:16,572 prompts, or to make our— to 291 00:18:16,636 --> 00:18:20,091 make— to get suggestions how to make process easier. 292 00:18:20,493 --> 00:18:23,482 But because industries want to maintain the 293 00:18:23,563 --> 00:18:27,197 confidentiality and maintain AI governance, 294 00:18:28,640 --> 00:18:32,160 so mostly AI won't affect that 295 00:18:32,224 --> 00:18:36,067 much. What I have seen, because mostly 296 00:18:36,487 --> 00:18:39,937 I have seen organizations who are in 297 00:18:40,114 --> 00:18:43,563 healthcare industries, they refrain from 298 00:18:43,820 --> 00:18:47,670 using AIs. So It 299 00:18:47,702 --> 00:18:51,543 really, like, I know AI can help you to 300 00:18:51,655 --> 00:18:55,190 optimize SQL queries, AI can help you 301 00:18:55,480 --> 00:18:59,219 build metadata and ETL pipelines and suggest a better 302 00:18:59,219 --> 00:19:02,796 architecture, but there is still a 303 00:19:03,039 --> 00:19:06,343 lot of issues with AI governance and data 304 00:19:06,825 --> 00:19:10,057 issues with the confidential and sensitive data which 305 00:19:10,524 --> 00:19:14,065 organizations own. That's a great 306 00:19:14,065 --> 00:19:17,833 segue because we talked about AI, we talked about that, but what 307 00:19:17,833 --> 00:19:21,358 parts of— how do you decide which parts of an ETL 308 00:19:22,037 --> 00:19:25,416 or ELT workflow are worth automating first? 309 00:19:25,834 --> 00:19:29,422 Like, how do you— is it about how often the job has to run? 310 00:19:29,760 --> 00:19:33,160 Is it about criticality? Is it something else, some mixture 311 00:19:33,553 --> 00:19:37,004 of the above? So mostly the most 312 00:19:37,004 --> 00:19:40,768 repetitive task which we perform should be 313 00:19:41,459 --> 00:19:45,172 automated using the tools we want. 314 00:19:45,638 --> 00:19:49,318 For example, every time transport— I would 315 00:19:49,367 --> 00:19:52,934 say, uh, whatever reports we need to 316 00:19:52,983 --> 00:19:56,390 generate, we first need to understand 317 00:19:56,567 --> 00:19:59,449 first the business requirements of the stakeholders, 318 00:20:00,190 --> 00:20:03,394 because, and because it is, every 319 00:20:03,909 --> 00:20:07,694 organization would have different business rules, and it becomes quite 320 00:20:07,694 --> 00:20:10,848 difficult for us to understand which 321 00:20:11,057 --> 00:20:14,357 records we should include, exclude while building a 322 00:20:15,016 --> 00:20:18,231 metric. So mostly, first is the 323 00:20:18,359 --> 00:20:21,782 business definition and the requirements of 324 00:20:21,815 --> 00:20:24,707 stakeholders, and secondly, on the 325 00:20:25,640 --> 00:20:29,337 base voluminous amount of data being generated, so you 326 00:20:29,467 --> 00:20:33,278 have to come with efficient design of a data architecture 327 00:20:33,930 --> 00:20:36,926 such that the data is maintainable and 328 00:20:36,974 --> 00:20:40,760 recoverable. For example, if the one of the data 329 00:20:40,905 --> 00:20:44,563 pipelines fails and you, you 330 00:20:44,691 --> 00:20:48,510 won't get the report being loaded in the table which 331 00:20:48,622 --> 00:20:52,136 we require for the reporting, it's better to show the 332 00:20:52,136 --> 00:20:55,312 last refresh historical data to the 333 00:20:55,360 --> 00:20:58,858 stakeholders instead of showing no data because 334 00:20:58,970 --> 00:21:01,997 the data failed, so you can gain 335 00:21:02,835 --> 00:21:06,315 stakeholders' trust. And most importantly, it 336 00:21:06,959 --> 00:21:10,652 is important to maintain a control log, 337 00:21:11,363 --> 00:21:15,171 like when the data is inserted or deleted, 338 00:21:16,012 --> 00:21:19,732 in what amount of duration data is getting inserted, and how 339 00:21:19,813 --> 00:21:23,468 many records were inserted in the database. So 340 00:21:23,615 --> 00:21:27,286 this is— and you even have to see that if there is any 341 00:21:27,449 --> 00:21:31,112 failure in the system, You have to monitor when you rerun the 342 00:21:31,659 --> 00:21:34,633 data pipeline, you have to see the last 343 00:21:35,211 --> 00:21:38,758 log point, like last timestamp when it 344 00:21:39,017 --> 00:21:42,773 did stop running, and you have to— when 345 00:21:42,886 --> 00:21:46,124 you start try to rerun the data, you have to see there is no kind 346 00:21:46,124 --> 00:21:49,954 of duplication happening in the data. So you have to 347 00:21:50,505 --> 00:21:54,221 maintain the idempotency of the data, so we 348 00:21:54,237 --> 00:21:56,781 don't have duplicate records while rerunning the data. 349 00:21:58,355 --> 00:22:01,089 And mostly, so this 350 00:22:02,658 --> 00:22:05,942 way, uh, you have to, you have to even 351 00:22:06,120 --> 00:22:09,420 maintain the valid or invalid record 352 00:22:09,938 --> 00:22:13,545 history because in future if business requirements change, 353 00:22:14,063 --> 00:22:17,762 it is really important to know, figure 354 00:22:17,842 --> 00:22:21,376 out in the data that what is, 355 00:22:21,666 --> 00:22:25,425 it is really important about what kind of data records are getting affected. 356 00:22:29,009 --> 00:22:32,636 Right. It's just interesting. Obviously, reports 357 00:22:32,748 --> 00:22:36,107 have to be reliable and accurate. 358 00:22:37,649 --> 00:22:40,043 What do you— how do you 359 00:22:40,959 --> 00:22:44,674 balance moving quickly with reliable reports? 360 00:22:45,256 --> 00:22:48,732 It seems like, you know, the old, you can't have it good— you can have 361 00:22:48,732 --> 00:22:52,070 it good, fast, or cheap, and you have to pick 2. Right. 362 00:22:52,619 --> 00:22:53,509 What's your take on that? 363 00:22:56,257 --> 00:22:59,992 For having reliable reports, first I would say 364 00:23:00,445 --> 00:23:04,025 I will provide an example. For example, I'm trying to 365 00:23:04,334 --> 00:23:07,771 calculate the percentage of 366 00:23:07,788 --> 00:23:10,995 completion rate of a visit. I will take the 367 00:23:10,995 --> 00:23:14,554 completion rate and a denominator as the total number of 368 00:23:14,618 --> 00:23:18,290 visits. So I don't know in total number of 369 00:23:18,322 --> 00:23:21,973 visits whether should I consider the invalid records 370 00:23:22,791 --> 00:23:26,467 and incomplete records of visits which were never done. 371 00:23:26,900 --> 00:23:30,415 So it is something which could be defined by a 372 00:23:30,431 --> 00:23:34,090 business. So we need to have the answers 373 00:23:34,523 --> 00:23:38,070 documented from the stakeholders, what kind of— what 374 00:23:38,440 --> 00:23:42,147 we need to include in the total visit denominator so we 375 00:23:42,147 --> 00:23:45,919 can get accurate number 376 00:23:46,079 --> 00:23:49,490 in our Percentage of completion rate. 377 00:23:49,973 --> 00:23:53,451 So this way we could have a reliable report which could 378 00:23:53,451 --> 00:23:56,729 show us accurate values, because 379 00:23:56,745 --> 00:24:00,420 sometimes the— our report to be technically correct, we 380 00:24:00,550 --> 00:24:04,257 have built a good SQL, we are doing the good 381 00:24:04,370 --> 00:24:07,412 data source extraction and 382 00:24:08,092 --> 00:24:11,506 validation and maintaining all the control logs, 383 00:24:12,121 --> 00:24:15,720 but Still technically the process is correct, but 384 00:24:15,720 --> 00:24:19,265 our reports will still be showing inaccurate values. 385 00:24:19,848 --> 00:24:22,709 So it is really important to have clear 386 00:24:23,661 --> 00:24:27,501 business requirements set. And secondly, you should have— 387 00:24:27,921 --> 00:24:31,470 you need to build the stakeholder trust by showing the 388 00:24:31,486 --> 00:24:35,309 last refresh rate in your reports. So 389 00:24:35,422 --> 00:24:39,242 it is a— this is where you can help their health building 390 00:24:39,274 --> 00:24:42,533 a reliable report. So it is 391 00:24:42,710 --> 00:24:46,515 like they can see the last refresh date, date, and if 392 00:24:46,644 --> 00:24:50,000 it is, it is past midnight and today it is 393 00:24:50,096 --> 00:24:53,918 daytime because the pipeline failed, but 394 00:24:53,966 --> 00:24:57,627 they would be like, yes, we have accurate data for which 395 00:24:57,964 --> 00:25:01,207 we want to make the decision. So this way we could make 396 00:25:01,207 --> 00:25:05,048 reliable reports. Taken 397 00:25:05,627 --> 00:25:08,167 into consideration these practices. 398 00:25:09,887 --> 00:25:12,427 Interesting. I often find 399 00:25:13,728 --> 00:25:15,400 it fascinating that 400 00:25:18,003 --> 00:25:21,667 people think that AI can do all the— all sorts of things, and 401 00:25:22,969 --> 00:25:26,089 I think ultimately human judgment, and people rely on these 402 00:25:26,427 --> 00:25:29,773 reports, right? Data engineers, data types, 403 00:25:30,482 --> 00:25:33,135 so to speak. When I was doing consulting 404 00:25:34,193 --> 00:25:37,415 back in the day, people hated doing reports, 405 00:25:37,816 --> 00:25:41,599 right? Probably because Crystal Reports was this awful software. And 406 00:25:41,599 --> 00:25:44,404 if you've never heard of it, you're very thankful to be born when you were. 407 00:25:45,014 --> 00:25:48,620 But I remember just having to work with it. 408 00:25:48,701 --> 00:25:52,372 Again, it wasn't really a fun experience, but the fact 409 00:25:52,404 --> 00:25:56,219 that it was able to— the fact that you 410 00:25:56,660 --> 00:26:00,472 You could. What really matters with reports is 411 00:26:01,050 --> 00:26:04,373 reports are what people who write the checks see, 412 00:26:04,758 --> 00:26:08,402 right? Whether that's a KPI dashboard or anything like that, the people who 413 00:26:08,450 --> 00:26:12,287 are the business users are the ones that see the reports. 414 00:26:13,170 --> 00:26:16,364 And that is crucial in 415 00:26:17,359 --> 00:26:20,746 business decisions are made. And plus they're also the people signing the checks usually. 416 00:26:21,372 --> 00:26:24,935 So if you keep them happy, that that's good for your 417 00:26:25,064 --> 00:26:28,770 budget. What would be your advice to someone starting out? If you 418 00:26:28,898 --> 00:26:32,652 had to go back and tell, you know, your 11th, 12th 419 00:26:32,748 --> 00:26:36,406 year student, what would you say to yourself or someone 420 00:26:36,438 --> 00:26:39,615 else in that similar position? I would say 421 00:26:40,064 --> 00:26:43,914 first to believe in yourself, because when I 422 00:26:43,979 --> 00:26:47,652 was doing my master's, I was quite keen 423 00:26:47,829 --> 00:26:51,404 that I would get an internship. To get 424 00:26:51,534 --> 00:26:54,603 real hands-on experience before moving 425 00:26:54,701 --> 00:26:58,532 into a full-time opportunity after my 426 00:26:58,597 --> 00:27:01,838 graduation. So when I did my 427 00:27:01,935 --> 00:27:04,447 internship at a manufacturing company, 428 00:27:06,120 --> 00:27:09,549 for— at Balmain Richardson Foods, so 429 00:27:10,115 --> 00:27:12,655 mostly I was— my work was to 430 00:27:13,334 --> 00:27:17,034 automate Power BI reports for the manufacturing 431 00:27:17,210 --> 00:27:20,405 flow. So 432 00:27:20,678 --> 00:27:23,985 mostly I would say that I 433 00:27:24,354 --> 00:27:28,143 was doing a lot of things out of the blue, like from 434 00:27:28,143 --> 00:27:31,723 a beginner Power BI knowledge and foundational knowledge, 435 00:27:32,157 --> 00:27:35,207 I was jumped into for advanced Power BI 436 00:27:35,833 --> 00:27:39,606 things and real-life problem requirements. For 437 00:27:39,606 --> 00:27:42,640 example, I need the data was 438 00:27:43,459 --> 00:27:46,652 imported into Power BI using API calls. 439 00:27:47,406 --> 00:27:51,016 So it does not like that if you are using any other data 440 00:27:51,241 --> 00:27:54,916 source, it would collect that time zone. Because of the API calls, 441 00:27:55,012 --> 00:27:58,783 it was collecting a different time zone, the 442 00:27:58,799 --> 00:28:02,506 default UTC. And in manufacturing, you 443 00:28:02,506 --> 00:28:06,260 need the data to be visible to the operators and the 444 00:28:06,260 --> 00:28:10,079 supervisors, which is 24/7, and the data should be visible 445 00:28:10,260 --> 00:28:14,020 at midnight also. So out of the blue, 446 00:28:14,020 --> 00:28:17,840 the report was blank and there was nothing showing 447 00:28:17,840 --> 00:28:21,560 up in the Power BI report. Wow. So it was kind 448 00:28:21,560 --> 00:28:24,920 of— yeah, so it was like, 449 00:28:24,920 --> 00:28:28,620 I'm doing everything right, I'm not doing— I'm doing— not 450 00:28:28,620 --> 00:28:31,700 applying any kind of formula on the time zone, what's 451 00:28:31,700 --> 00:28:35,440 happening? And later I figured out that it 452 00:28:35,440 --> 00:28:39,240 is because Power BI defaults to UTC time zone and you 453 00:28:39,240 --> 00:28:43,014 don't have any settings in Power BI So you can change the time 454 00:28:43,014 --> 00:28:46,379 zone. So I have to develop 455 00:28:46,860 --> 00:28:50,289 that formula to change from UTC to 456 00:28:50,385 --> 00:28:54,007 EST or EDT time zone. So because of 457 00:28:54,071 --> 00:28:57,741 daylight savings, it was a bit difficult. And you have to 458 00:28:57,998 --> 00:29:01,731 work and you can't find a solution anywhere on the internet 459 00:29:01,908 --> 00:29:05,513 and you have to use your own knowledge to come up with the 460 00:29:05,610 --> 00:29:09,023 idea. So it was a big troubleshooting 461 00:29:09,359 --> 00:29:12,949 ability I gained in my internship. And secondly, 462 00:29:13,714 --> 00:29:17,486 I need to— like, most of the reports need to be deployed 463 00:29:17,567 --> 00:29:20,894 on it. Some of the reports need to deploy on TV so the 464 00:29:21,056 --> 00:29:24,891 workers on the floor can 465 00:29:25,246 --> 00:29:28,202 see the TV, what the Power BI reports, and know 466 00:29:29,735 --> 00:29:33,412 the required business, what is happening. But like, they could 467 00:29:33,606 --> 00:29:37,219 know about downtime of a machine, or they could know 468 00:29:37,316 --> 00:29:40,484 about overall hourly 469 00:29:40,516 --> 00:29:44,188 performance of what is happening in the 470 00:29:44,509 --> 00:29:48,213 manufacturing flow. So when the 471 00:29:48,213 --> 00:29:51,854 report is deployed on TV, you can't have user interaction. 472 00:29:51,966 --> 00:29:55,446 And I was like, the challenge was like how to 473 00:29:55,526 --> 00:29:58,605 make the Power BI refresh every— 474 00:29:59,840 --> 00:30:03,481 to refresh every 5 or 10 minutes or 15 475 00:30:03,609 --> 00:30:07,362 minutes without any user interaction. So it 476 00:30:07,362 --> 00:30:10,635 was a challenge for me too, because if you 477 00:30:10,747 --> 00:30:14,549 notice, when a Power BI report refreshes, there 478 00:30:14,677 --> 00:30:18,190 would be any kind of human interaction, or you would 479 00:30:18,190 --> 00:30:21,944 refresh the browser. So that was a major challenge, 480 00:30:22,120 --> 00:30:25,922 and I have to come up with some unique solution from my own way. 481 00:30:25,970 --> 00:30:28,681 Yeah, interesting. 482 00:30:29,676 --> 00:30:32,746 There's just no substitute for doing to learn, is there? 483 00:30:34,670 --> 00:30:37,145 No, those horrible experiences 484 00:30:39,441 --> 00:30:43,064 tend to stick with you and you learn the most from. I haven't 485 00:30:43,064 --> 00:30:46,364 figured out a way to learn with, uh, and still avoid them. 486 00:30:46,801 --> 00:30:50,165 If anyone else, you know, who's listening to this has found that, I would love 487 00:30:50,197 --> 00:30:53,938 to know about it. Audible is a sponsor. Do you listen to audiobooks 488 00:30:54,391 --> 00:30:57,888 or— Oh yes, I do 489 00:30:57,888 --> 00:31:01,124 listen to audiobooks. Any recommendations for our audience? 490 00:31:02,502 --> 00:31:06,248 I would say Storytelling with Data 491 00:31:06,329 --> 00:31:08,835 by Paul Nuttlick. Okay. 492 00:31:10,629 --> 00:31:13,959 Interesting. Very cool. Audible is a sponsor of Data Driven, 493 00:31:14,089 --> 00:31:17,562 and you can get yourself a free book if you're listening and you go to 494 00:31:18,110 --> 00:31:21,917 thedatadrivenbook.com. That routes you to Audible and you'll get a free audiobook 495 00:31:21,982 --> 00:31:25,596 on us, and we get a little tap on the shoulder that'll probably be enough 496 00:31:25,628 --> 00:31:29,346 to buy a cup of coffee. Maybe not, depending if it's Starbucks or 497 00:31:29,623 --> 00:31:33,458 gas station coffee. What's your favorite part of your current gig? 498 00:31:35,420 --> 00:31:38,841 My favorite part is when I have already built 499 00:31:38,841 --> 00:31:42,521 existing projects which are already deployed, 500 00:31:42,699 --> 00:31:46,522 and when somebody comes out, there is an error or there's 501 00:31:46,586 --> 00:31:49,907 some issue, I don't know, I enjoy to notice 502 00:31:50,681 --> 00:31:54,356 the discrepancy and Troubleshoot 503 00:31:54,533 --> 00:31:58,309 the issues. So that's my favorite 504 00:31:58,406 --> 00:32:02,166 part of gig. And the challenges I, I told you 505 00:32:02,618 --> 00:32:06,061 which I faced during my internship for the things 506 00:32:06,659 --> 00:32:10,313 which are out of the box and you can't find a solution anywhere 507 00:32:10,523 --> 00:32:13,404 on the internet, uh, that's nice. 508 00:32:14,385 --> 00:32:18,229 And I really like— I learned a lot of things from my job because 509 00:32:18,380 --> 00:32:21,970 First, I was a data analyst, and from 510 00:32:22,035 --> 00:32:24,933 beginner I jumped to advanced because of the 511 00:32:25,722 --> 00:32:29,520 things which were— because of the requirements which were given to me. 512 00:32:29,987 --> 00:32:33,220 And then later I became a data engineer in my 513 00:32:33,396 --> 00:32:37,036 job, so I learned various things— how 514 00:32:37,246 --> 00:32:40,246 to do link server and how to 515 00:32:40,649 --> 00:32:42,730 interact with multiple databases. 516 00:32:44,585 --> 00:32:47,928 And how to use OpenQuery, how to build 517 00:32:48,121 --> 00:32:51,930 effective data architecture based on a volume of data, how to 518 00:32:51,979 --> 00:32:55,402 manage 100 million amount of records. And 519 00:32:55,996 --> 00:32:59,564 because in Power BI you can't import 100 520 00:32:59,773 --> 00:33:03,604 million number of rows, you need to optimize before importing 521 00:33:03,636 --> 00:33:07,312 the data in Power BI. So you have to be creative 522 00:33:07,651 --> 00:33:10,853 around getting a good data architecture 523 00:33:11,792 --> 00:33:14,125 and streamline the process. 524 00:33:15,793 --> 00:33:19,433 So that's my favorite part of the work I do, 525 00:33:20,127 --> 00:33:23,825 what I learn every day in my job. Very cool. 526 00:33:24,745 --> 00:33:27,329 What do you think the coolest thing in technology is today? 527 00:33:29,210 --> 00:33:32,247 For me, I know many people would say AI, 528 00:33:32,799 --> 00:33:36,550 but I was working in a health industry You 529 00:33:37,066 --> 00:33:40,554 help mostly organization refrain from using AI 530 00:33:40,715 --> 00:33:44,551 for the sensitivity and confidential data. So for me, 531 00:33:44,760 --> 00:33:48,223 automating the reports using Power BI, 532 00:33:48,400 --> 00:33:51,670 Python, and SQL is the most interesting 533 00:33:52,104 --> 00:33:55,775 part in the technology than AI. I can just 534 00:33:55,986 --> 00:33:59,777 get— use AI prompts to get suggestions how to make 535 00:33:59,777 --> 00:34:03,163 the system more effective and how to make 536 00:34:03,533 --> 00:34:06,042 each of the reporting 537 00:34:06,653 --> 00:34:09,967 lifecycle of data easier. 538 00:34:11,230 --> 00:34:14,363 But I would say reporting 539 00:34:14,460 --> 00:34:18,108 automation is the most— using SQL, 540 00:34:18,237 --> 00:34:21,481 Power BI, and Python is the 541 00:34:21,628 --> 00:34:24,538 most— is the thing that fascinates me in the technology. 542 00:34:26,247 --> 00:34:29,655 Interesting. What are you looking forward to technology doing one day? 543 00:34:30,851 --> 00:34:34,100 So I, I look 544 00:34:34,327 --> 00:34:38,162 forward that we don't have to do repetitive 545 00:34:38,274 --> 00:34:41,576 tasks and have all the 546 00:34:41,576 --> 00:34:45,300 reports automated and just work 547 00:34:45,365 --> 00:34:48,805 on the maintenance of the reports rather than 548 00:34:48,950 --> 00:34:52,196 still a lot of— I see a lot of 549 00:34:52,921 --> 00:34:56,184 reports are being generated and people being very 550 00:34:56,620 --> 00:35:00,029 like, they are very, uh, like used to, or like for 551 00:35:00,029 --> 00:35:03,610 example, when I was doing reporting automation in 552 00:35:04,191 --> 00:35:07,919 manufacturing for manufacturing flow, though the reports 553 00:35:07,919 --> 00:35:11,627 were automated, people were still using paper timesheets 554 00:35:11,708 --> 00:35:15,344 to validate it. They didn't rely on it. So 555 00:35:15,714 --> 00:35:17,709 I would say mostly 556 00:35:19,410 --> 00:35:22,569 build a more reliable reporting automation 557 00:35:23,163 --> 00:35:26,423 where people can trust. Interesting. 558 00:35:27,900 --> 00:35:31,737 Where can folks find out more about you and what you're 559 00:35:31,753 --> 00:35:35,495 up to? People can find more about 560 00:35:35,623 --> 00:35:39,091 me on LinkedIn. So you can see 561 00:35:39,461 --> 00:35:42,479 articles written by me and the 562 00:35:42,608 --> 00:35:46,373 podcast I participated in, and 563 00:35:46,373 --> 00:35:50,014 the— in the roundups I 564 00:35:50,336 --> 00:35:52,882 participated in. And you can 565 00:35:53,966 --> 00:35:57,589 know more, you can get the articles and roundups and the 566 00:35:57,589 --> 00:36:01,374 opportunities I go for and learn 567 00:36:01,471 --> 00:36:03,247 more about me mostly through LinkedIn. 568 00:36:05,296 --> 00:36:07,656 Interesting. Very cool. Any parting thoughts? 569 00:36:09,683 --> 00:36:13,115 What do you say? So any, any final thoughts that you'd like to share with 570 00:36:13,115 --> 00:36:16,759 the audience? The final thought I would share 571 00:36:16,888 --> 00:36:20,388 with the audience is that stay curious and 572 00:36:20,500 --> 00:36:23,599 don't stop learning. And if you want to learn something, 573 00:36:24,113 --> 00:36:27,725 try to, try to participate in, uh, if 574 00:36:27,789 --> 00:36:31,594 you're a beginner, try to participate in hackathon because it 575 00:36:31,723 --> 00:36:35,399 could change the way you look at things and you will 576 00:36:36,112 --> 00:36:39,857 network with fascinating people early in your 577 00:36:39,857 --> 00:36:43,311 career. And if you are a beginner, definitely 578 00:36:43,733 --> 00:36:47,373 go for internship because during internships employers are a 579 00:36:47,373 --> 00:36:50,110 bit lenient and they just want to know how 580 00:36:50,689 --> 00:36:54,264 excited and what— you have good amount of knowledge 581 00:36:54,441 --> 00:36:58,063 before applying the hands-on skills. But if you are going for a 582 00:36:58,079 --> 00:37:01,799 full-time opportunity, it becomes really important that 583 00:37:01,799 --> 00:37:05,510 you have close industry experience because people don't want to train 584 00:37:05,510 --> 00:37:08,910 you. So always keep learning. And while, uh, in my 585 00:37:08,910 --> 00:37:12,730 job mostly I learned a lot of things. And when 586 00:37:12,730 --> 00:37:16,430 you look at a legacy architecture or listen 587 00:37:16,430 --> 00:37:19,390 to people who are already in the industry, 588 00:37:20,500 --> 00:37:24,070 you— in— when I go for my speaking opportunities, you 589 00:37:24,070 --> 00:37:27,510 come up with different solutions. It is not that only that 590 00:37:27,510 --> 00:37:31,110 non-STEM people only have to come with creative 591 00:37:31,110 --> 00:37:34,676 ideas. STEM people even have to be 592 00:37:34,676 --> 00:37:38,498 creative with their problems which we are trying to solve. And it's 593 00:37:38,498 --> 00:37:42,239 quite fascinating to see that even 594 00:37:42,239 --> 00:37:45,660 in STEM, each problem would have different solution 595 00:37:46,960 --> 00:37:50,317 and effective way in doing it. So always 596 00:37:50,590 --> 00:37:54,363 try to learn and be curious. Very cool. What a great 597 00:37:54,363 --> 00:37:57,944 note to end on. Thanks for joining. Thanks for being on the show. We'd love 598 00:37:57,976 --> 00:38:01,399 to have you back again. You can Go really deep into the nerdy 599 00:38:02,075 --> 00:38:05,838 Power BI and SQL Server stuff with Andy. And for this, we'll let the 600 00:38:05,838 --> 00:38:08,314 credits roll. Awesome. Thank you very much.