From IIT Kanpur to The LLM Revolution: Meet Kognitos Engineer Siddarth Agrawal

Can you tell a little bit about your life growing up?

My father was a railway officer so my brother and I got to study in a lot of different schools (senior government officers get transferred a lot here in India). We had a pretty great time getting to know the different cultures of various Indian cities. In my high school, we had a computer science course, and they taught us how to program in C – basic stuff like “hello world” programs or the “create those weird patterns using for loops”. But I was completely captivated by this methodical and systematic approach to solving problems. I really liked how a computer language was so logical and mathematical (as opposed to the English language which I was admittedly not that great at at the time). I knew then and there that I had to pursue a career in Computer Science.

You attended the Indian Institute of Technology (IIT), a prestigious school for computer science. What did you learn at IIT?

The professors at IIT focused a lot on the fundamental aspects of computer science. These building blocks make it easier for me to reason why a system is failing or an application behaves the way it does. It helps with architecture, scale, and all sorts of problems. It gave me the tools to approach any problem.

After IIT you’ve worked at a few different companies, including Nutanix and AI Dash, two very different companies. What did you learn from each?

I did my college internship at Nutanix, and then joined the platform team. Nutanix was a great starting point for me and helped me get my legs underneath me. I learned how coding in companies is very different from coding on your own or in college. It taught me how to write code properly and how to manage a project. It taught me what being an Enterprise engineer is all about. But after time I found that my learning had slowed and I wanted to be at a smaller company where I could move faster.

There’s a trade-off isn’t there? You want to build things quickly and experiment to produce innovative products, but at the same time need to build things that will scale and deliver consistent experience. How do you balance this?

Yes there is always a tradeoff and my experience at AI Dash gave me insight into the other side of the coin. I joined as the 24th employee. The culture at AI Dash was very different. They had a “Get Sh*t Done” culture. This works for a while but then can put the platform under a lot of stress over time. Relative to Nutanix I found this culture the complete opposite. So there is a tradeoff between speed and quality at times. Finding top engineers/talent and being a part of a team that helps improve each other is the only way to move quickly while still building something properly. 

So how did you find Kognitos?

Everyone knew Binny as a superstar at Nutanix even though I didn’t know him personally. We are also both alumni of IIT Kanpur. One day I saw he posted a video of processing documents completely in English and I quickly realized the potential for LLMs so I cold emailed him and the rest is history!

Wow! This is interesting. As you know we are hiring for a lot of roles right now, what advice would you give to engineers interested in these roles?

First you have to establish a connection. Binny and I went to the same college and had the same company background. Next you have to establish experience, and clearly say why your experience helps. Lastly, you must connect with the vision and why you can help with that vision. It has to be short, crisp and clear.

You’ve been at Kognitos for a year now. What about the team or culture that you really like and what are the skill sets needed to be a good fit?

I wanted to join a team with more experienced engineers from whom I could learn. People here know how to build products that scale into multi-Billion dollar companies. Kognitos is the type of environment and where everyone challenges each other and themselves so we all improve. We set extremely high standards for our code. It has to be A grade. We have a very robust test framework and so your code has to live up to the high bar.

Only ambitious people can thrive here. We have conflict to make us all better. You have to give more than 100% each and every day and that might take a toll on some people, but that is the intensity we want to set and if you enjoy building software you will love it. You can be a part of something that will truly impact people at scale. I wanted to build something where even my parents will be impacted in how they use computers and Kognitos is that type of product.

There is a lot of buzz around Generative AI. What has you excited about this category and Kognitos today?

The possibility of LLMs, Generative AI and Kognitos is massive. We are democratizing things to where even people who have very little experience with computers can interact with computers directly using plain language. Today it is just in plain English but in the future in other languages, Tamil, Hindi, Spanish etc. I’m helping build something that will be a part of an extraordinary revolution in computer science, and it’s exciting to know that I can leave behind my name as an integral part of that legacy!.

What do you think about Generative AI being deployed in a safe and consistent way for the Enterprise? What has to be done?

In my experience, people don’t really know what’s going on in their business. They often have a very incomplete picture, especially if there is variation in a process.. With Kognitos we can fill in that gap, provide that information and help companies capture not only how their business operates today, but give their employees the power to automate away parts of this work that are manual. Businesses also need the capability to audit LLM outputs and know why something went wrong (and things usually will go wrong in any enterprise system). Our architecture and design make this possible through Conversational Exception handling and the ability to audit in English. This is the right approach for the enterprise.