Analytics Vidhya has launched its new sequence, ‘Main With Knowledge,’ the place trade leaders share their experiences, profession journeys, attention-grabbing tasks, and extra. Within the debut episode of the sequence, Mr. Anand S., CEO and Chief Knowledge Scientist at Gramener, was interviewed by the Founder & CEO of Analytics Vidhya, Mr. Kunal Jain. Anand is a visionary within the discipline of Knowledge Science and a frequent speaker at DataHack Summit. He’s recognized for the thrilling tasks he works on and the best way he narrates tales with detailed visualizations. In this interview, Anand unveils the magic he brings to his talks, sharing with us a glimpse of his storytelling expertise and the thought course of behind them. Listed below are among the considerate insights from the interview.
The Artwork of Storytelling with Anand S.
Kunal J: Each time I attend your speak, I simply stroll out awestruck by the way you narrate the story and the visualizations. Are you able to stroll us by way of what goes behind creating these talks and the way you strategy them?
Anand S: There are normally two sorts of talks that I ship – one aimed toward instructing and the opposite at studying. The training speak is the place I decide one thing I don’t know sufficient about, study it & then discuss it. It helps me be taught one thing new inside a set deadline. It additionally helps me discover out what materials works and what doesn’t. So, the following time I speak, I do know what to repeat and what to switch or convert to tales. The second type of speak is the place I goal to show by assembling what has labored properly in a lot of my earlier talks, selecting simply the attention-grabbing elements, and stitching them collectively.
Kunal J: I’m presuming lots of what you do within the talks the place you might be studying would finally go into the instructing talks. So how precisely does the precise creation or the analysis part for a studying speak occur?
Anand S: I discover purpose-driven studying works higher than systematic studying. What I imply by that’s if I picked up a ebook or an API’s website and went by way of it finish to finish. I be taught lower than if I decide up an issue and be taught to resolve it. I additionally discover that if I decide an issue that I’m concerned with, I’m extra prone to get an answer than an issue that I’m simply making up. So, the primary process is to determine an issue that I wish to clear up. The second half is knowing the issue and attempting to resolve it, which is an entire lot simpler as a result of I’ve one thing that I’m concerned with.
The Influence of Storytelling
Kunal J: The tales you share in your talks are very attention-grabbing. What obtained you into that format of presenting? Do you discover it to be extra impactful?
Anand S: Individuals have a tendency to recollect the tales informed in a chat extra typically than the remainder of what is alleged. So, it’s more practical to convey the content material in a narrative format. Whereas the content material of the story is what issues, I’ve realized it’s the means these tales are informed that makes them memorable.
For instance, after I would inform people who older kids have a tendency to attain extra, they’d say I may have guessed that. So I modified the title to say, ‘Do you suppose the solar signal impacts the marks of a kid?’ At which level they started to suppose, ‘No, that may’t be.’ And the truth is, sure, it does – which makes it a hell of much more memorable. Once I speak in regards to the community of the method of Characters within the Mahabharata, it’s attention-grabbing. But it surely’s solely after I began narrating it as, ‘Let’s discover out which Pandava was Draupadi’s favourite,’ that it began turning into extra memorable. And due to this fact, the way you inform it clearly has a robust influence.
Now, what’s attention-grabbing in a chat might not essentially be content-related. To provide you an instance, I used to be exploring a chat on the usage of comedian characters. It struck me that whereas there are various attention-grabbing issues about automating comedian characters, one explicit story significantly resonated each with me and the viewers, which is that, as a child, I needed to attract cartoons for Disney. And the truth is I can’t; even should you put a gun to my head, I can’t draw. So I gave up. However as a result of I put collectively, together with my colleagues, this frequent technology API referred to as Comicgen, utilizing their drawing expertise mixed with my programming expertise – which occur for use at Star TV, which obtained acquired by Fox, which then obtained acquired by Disney – I ended up drawing for Disney, in a really oblique means.
Now, right here’s the factor. That is an attention-grabbing story however has completely nothing to do with the know-how behind or the info behind the speak itself. And that’s what I imply after I say it’s not that the storyline essentially needs to be one thing that strengthens the content material immediately. It may be one thing that augments the content material otherwise or is a narrative round it. So I attempt these out, and whichever works will get distilled after which will get into the speak.
Speaking to the Lots vs. Speaking to Purchasers
Kunal J: How is the storytelling in your public talks completely different from the talks you might have together with your purchasers? Does the method stay the identical, or is it a really completely different setup?
Anand S: It’s 50 % completely different. There are purchasers for whom we do nearly precisely the identical type of work. It’s simply that it’s for personal information. So, when an organization involves us with an issue, I might narrate to them a associated real-life downside with their information. It could be offered in some type of gross sales storytelling format that’ll keep memorable. After which, we’d convert it into an utility that delivers the identical story.
However that’s solely 50 % of what we do with purchasers as a result of the opposite 50 % is, whereas in a chat, folks count on a completed product. Whereas working with purchasers, they count on two issues which can be completely different. First, to be concerned as a part of the method. So there’s lots of working course of output that goes alongside. Second, it’s the potential for them to have the ability to ship such insights. So typically, they’d say, “Give me the uncooked materials for me to have the ability to mess around. I don’t need you to give you the reply. I would like to have the ability to give you the reply myself. So on the subject of that, what we do is extra instrument constructing or answer constructing that empowers them to seek out the story and narrate the story. In that sense, it’s completely different.
Anand S.’ Tackle Generative AI
Kunal J: As somebody who’s actually nice at constructing visualizations and information tales, how have your experiments with generative AI been, and the way do you are feeling about it?
AI lowers the price of doing sure issues, and when one thing has a decrease price, much more issues change into price doing.
Anand S: The final time I felt that the world had modified round me this a lot was when Google was launched. That’s when google.stanford.edu got here out. This principally meant that I by no means wanted to be taught stuff. I may lookup something. And with LLMs coming in, now I understand I by no means must suppose; I simply must learn to get issues to suppose for me. It’s that transformational.
LLMs, I might say, are making the unimaginable potential. Let me let you know how. A process that may have taken me 10 hours earlier than will take me only one hour now, due to LLMs. That saves me 9 hours, which I can put into different issues, and that’s the distinction between unimaginable and potential. So now, we get higher high quality output with decrease effort.
How do I see it altering the best way I work? By and huge, I’ve stopped coding. I simply inform LLMs to code for me, and there have been just a few functions that I’ve constructed which can be 100% LLM generated. In different phrases, I didn’t write a single line of code, and dissipation works fantastically. Generative AI, particularly massive language fashions, has given me the power to create assistants. I can now create an assistant who will code for me or create concepts for me. I can create an assistant for any of a sequence of various makes use of. Now that I wish to be taught by experimenting, I exploit it as a self-help coach.
If that is so, I’m solely restricted by my creativeness within the type of assistants I want. That is significantly an issue for me as a result of I’m the type of one that doesn’t ask for assist. Even when I’m stranded and misplaced in a desert, I received’t ask for instructions. It’s the change that I’m attempting to get to. In 2000 or 1998, when google.stanford.edu was launched, I made a decision that I wanted to make a transition from understanding to with the ability to uncover. Now, I must make a transition from considering to with the ability to information considering. It’s going to be highly effective.
Highlights of Anand S.’s 2023
Kunal J: I’m a subscriber of your weblog, and I get this annual reflection that you simply ship out, itemizing what you’ve achieved previously 12 months and what you want to do that 12 months. When did that begin? Additionally, may you please share among the high matters you might be specializing in this 12 months?
Anand S: Properly, I began this about 4 or 5 years in the past. The principle intent was for me simply to maintain myself trustworthy in regards to the issues that I needed to do this 12 months and share what I’ve discovered within the course of. Publicly committing to attaining one thing retains me pushed, understanding I’ve to replace them about it the following 12 months.
This 12 months, I’ve taken on 3 issues publicly and a few issues privately. One is to run 50 experiments and be taught from them. I had executed this final 12 months as properly, nevertheless it began after which fizzled out. Once more, this 12 months, it began properly after which fizzled out. So now I’m attempting to construct a behavior of studying from experimentation.
One other, which is approaching slightly higher, is altering my setting to see the way it adjustments me in flip. So I attempt issues like figuring out of a special place day-after-day, assembly a brand new particular person day-after-day, listening to a special style of music day-after-day, and many others. The plan was each month, I might run certainly one of these adjustments in experiments. And a few of these have labored dramatically, whereas others haven’t. It’s a broad course of the place I decide to sharing and utilizing that suggestions to enhance.
The third factor I’ve taken up is calendar integrity, which is principally sticking to what I do. That is one thing that I used to do fairly properly. But final 12 months, I felt I slipped slightly bit and took it up this 12 months. But it surely’s extra like a behavior; when you’ve executed it, you don’t actually neglect.
Some Parting Recommendation
Kunal J: Based mostly in your perspective and expertise gained over time, what could be your high recommendation for somebody beginning off their profession immediately? Or, in different phrases, what would you advise your personal youthful self immediately?
Prioritize the issues that you simply like first. Merely since you discover it simpler to stay to.
Anand S: The 2 are considerably completely different. So, first, let me share what I might inform somebody beginning of their profession. The primary piece of recommendation could be to seek out and clarify what you want. It makes work so much simpler once you’re engaged on what you want. The second factor is to be taught by instructing. When you want to be taught one thing, one of the best ways to do it’s by committing to show it to someone. That has 3 advantages: First, it pressures you to be taught. Second, it improves the standard of studying to the purpose the place you will get someone else to clarify. Third, you’ll make some buddies who owe you a favor.
Now, if I needed to advise myself, going again a technology, then I’d most likely inform myself to not fuss about what’s good or what’s proper. Each single factor has an excellent and a foul – simply consider each. I say this as a result of I at all times have, and nonetheless do, discover a solution to a query with out acknowledging that a solution is only one means of taking a look at a query.
These had been the highlights from our unique interview with Gramener CEO, Mr. Anand S. You’ll be able to watch the complete interview right here. Keep tuned to our ‘Main With Knowledge’ sequence on the Analytics Vidhya Group Platform for extra unique interviews.