Employee Spotlight #5: Hadayat Seddiqi, Former Director of Machine Learning At Ontra

Are you a founder looking to learn from other founders on how to build your startup? Are you someone interested in joining a startup as an early-stage employee? Subscribe?to my mailing list, Founder to Founder (F2F):?f2f.substack.com.?

• Startup Spotlights are a high-level overview of a company that comes out once a week every Monday for free subscribers.
• Case Studies are in-depth research profiles on startups for paid subscribers. A paid subscription is 7$/month or $60/year. Highly recommended if you're looking to join an early-stage startup that you want more information on.

Check out my latest F2F stories:?

• Startup Spotlight #183: Candor - Candor provides you with a professional profile to share how you really work.
• Case Study Series: Here's How AudioMob Solved Its Growing Pains As It Evolved From An Early-Stage Startup To A Growth-Stage Company

Hadayat Seddiqi is a scientist and engineer working in machine learning. He has a background in physics research, most notably in quantum computing, where he worked on simulation and machine learning algorithms. After a stint at NASA working on space exploration robots, he moved to industry, where he built software and algorithms for a next-gen DNA sequencer. Following that, he built the AI function for understanding contracts as an early employee through hyper-growth at Ontra. Seddiqi enjoys reading about history, economics, and science fiction in his spare time and likes climbing rocks and running long distances with his dog Denali.

Note: Seddiqi is a former director of machine learning at Ontra.

Ontra's Director of Machine Learning Hadayat Seddiqi.

Frederick Daso: What was your journey to becoming a Director of Machine Learning at Ontra?

Hadayat Seddiqi: I was an early engineer when the company was around 15 people. The other three engineers were working on the app, and I was tasked to build a system that could understand contracts. A couple of years in, we were ready to add more people after some initial progress, so my work began to include management as well as technology development. After close to 5 years of incredible growth and a world-class valuation, we’ve built a great team with strong collaborations and a super efficient engine for delivering machine learning for our analytics and product. Machine learning is still very new, and contracts understanding?is an enormously complex problem. I was grateful to work with leaders who understood that advanced tech requires many iterations, often has unique resource needs, and takes a lot of time and patience to develop.

Daso: There’s a lot of conventional career advice about being a successful machine learning director, but are there any unorthodox lessons you’ve learned through experience or mentorship that more of your fellow machine learning directors or engineers should know?

Seddiqi: There’s a lot said about good leadership, building good culture, and managing your own psychology as a leader. Motivational energy is the most important thing for a team’s success, so engineering the culture to build and maintain that is paramount. Developing advanced technology requires a lot of humility and grit, so the culture needs to counter a reality filled with many failed experiments. The way you’re feeling is passed on too, so it’s important to stay positive and healthy for yourself and your team.

Some less orthodox advice: insights fuel scientific progress. Put in serious effort to convince people, especially your executive team, that forging relationships with domain experts is critical to success. Work to strengthen those collaborations over time until they’re tightly integrated into your operations.

Finally, seek to build an iterative and collaborative mindset within the team. It does no good for a science project to fail without learning something. And it certainly does no good for projects to go on for too long. Break it down, find two-for-one opportunities, and iterate your way to success. Stress this on the infrastructure side too. It is a myth that the uncertainty of science or the long-term view of infrastructure means they can’t be highly iterative projects that continuously deliver value.

Daso: What’s the toughest project (professionally or personally) that you worked on as a Director in Machine Learning or in general? What were the most important lessons you learned from that project?

Seddiqi: One of the hardest projects I worked on required subject-matter experts in contracts (lawyers). Lawyers happen to be very busy people, and they have a very different background than engineers or scientists that work with data. There is commonality somewhere, and the challenge is to pierce the language barriers to make progress together. Over time the lawyers began to understand data modeling, and my team and I understood how contracts worked.

In the beginning, we tried to solve problems ourselves, and we just hit wall after wall. It took a lot of time and energy but forging those collaborative relationships towards a deeper knowledge of the problem paid off in spades. I never imagined what a lawyer told me could influence architectural decisions in ML infrastructure, but it most certainly did.

(Collaboration on complex problems with domain experts is still radically challenging in many other fields, and it fundamentally bottlenecks progress with machine learning. Solving this problem more broadly happens to be my current focus.)

Daso: Who are some of the most inspirational people you’ve gotten to work with during your tech career?

Seddiqi: I’ll mention two people from my early days at Ontra: Lane Lillquist, our CTO, and Rich Niles, a now-retired engineer. Lane was a model for building a great culture, from encouraging us to deeply understand our recruits during interviews to making investments that boosted our productivity. I saw his focus on making Ontra a great place to work for engineers pay off enormously (for a while, we had a 100% retention rate!). Thank you, Lane, for taking post-lunch naps so that I could too.

Rich is one of those people who knows a lot and has experienced a lot. I remember the early days pair programming and learning so much about the history of computing, how he built the first products that he later sold to Microsoft, how he worked on group theory before it was cool and how his work is referenced in undergrad textbooks now. More than anything, he was just always a positive presence and sharp mind. Thank you, Rich, for educating me on the origins of \r\n.

Daso: How would you define your company’s culture, and how does it create an environment where you can do your best work?

Seddiqi: Everyone at Ontra prioritized working with genuine people who are highly motivated. Happiness at work for me was being surrounded by people you like and working towards an ambitious goal. I saw this fuel a lot of success in the company.

Daso: What’s one interesting thing (non-work related) that more people should know about you?

Seddiqi: I like cardio-heavy sports like soccer, snowboarding, and running. I live in Los Angeles, which is close to Joshua Tree, and I go there several times a year to do some scrambling. It’s fun, a little dangerous if you’re moving quickly, and Joshua Tree is just other-worldly, so it feeds my connection and appreciation for nature.

Daso: What’s something you want to accomplish in your career that you haven’t yet? What motivates you to get there?

Seddiqi: I mostly think about problems I’d like to solve. My background is in science, and I think it’s important we expand our scientific capabilities. Companies that can do science at a large scale have an extraordinary impact, and having done that once at Ontra, I’d like to open that up to many companies.

At some point, I’d like to take my learnings from the digital world to the physical world. I think it’s important that people continue to do harder things as they’re met with success. Building technology in the physical world is hard (I experienced this first-hand during my work in robotics and biotech). Certainly, energy, climate, biotech, and manufacturing are interesting industries, but it’s hard to say what impactful problems will exist by the time I get there.

Daso: Name three other individuals in the tech industry I should interview.

Seddiqi:?Here are my recommendations!

• Humayun Irshad, Engineering Manager of Machine Learning at Pano
• Justin Selig, Product Manager at Cerebras Systems
• Kieran Junge, CEO at GreenCubes

Are you a founder looking to learn from other founders on how to build your startup? Are you someone interested in joining a startup as an early-stage employee? Subscribe?to my mailing list, Founder to Founder (F2F):?f2f.substack.com.?

• Startup Spotlights are a high-level overview of a company that comes out once a week every Monday for free subscribers.
• Case Studies are in-depth research profiles on startups for paid subscribers. A paid subscription is 7$/month or $60/year. Highly recommended if you're looking to join an early-stage startup that you want more information on.

Check out my latest F2F stories:?

• Startup Spotlight #183: Candor - Candor provides you with a professional profile to share how you really work.
• Case Study Series: Here's How AudioMob Solved Its Growing Pains As It Evolved From An Early-Stage Startup To A Growth-Stage Company