Speaker Spotlight – Arjun Manjunatha

My story starts with two ventures I built in my undergrad. The first one was an app. During my undergrad, we had a long queue in the university cafeteria. We were spending a lot of time waiting for our food. So I built an app that would allow students to pre-order food and collect the food in the cafeteria rather than wait in a queue. This gained traction and had more than 100 users.

The next thing I built was a desktop application that would take in the websites to be blocked for a specific time and deny access to the user for those websites for a specific time period. This helped people focus on work and avoid distractions.

After this, I started my master’s in computer science at the University of Leeds. For my thesis, I worked on regression models to predict the spatial navigational ability of people from different socioeconomic backgrounds.

How did you start out in your tech career?

I did my undergrad in computer science because society told me this career pays well. But during my undergrad, I was introduced to Machine Learning and I was fascinated by the idea of machines being intelligent. So I went and did a master’s from Leeds and moved to London for opportunities. I joined a startup called Sention, and I’m still working here. That story is not that exciting, but that’s pretty much how I got into Machine Learning.

What are the signs of success in your field?

For me, having the confidence to build things is a sign of success. Another indicator is seeing someone use what I’ve built and hearing them say positive things about it. If they provide negative feedback, I don’t see it as a failure but rather as guidance on the path to achieving success. As cheesy as it sounds, this mindset has helped me focus on the long term rather than worrying about mistakes that have a short-term impact.

What is the best and worst thing about your job role?

For me, building models is the best part of my job. I enjoy brainstorming new ideas and implementing them. For example, in a project where we needed to read the barcode on a battery, we used YOLO models to detect the battery and barcode, and then used another Python library to read the barcode. The process of solving each challenge and then moving on to the next one is incredibly satisfying to me. I don’t enjoy cleaning data, but I understand that it is the most crucial part of any machine learning project.

What can you advise someone just starting out to be successful?

The best way to learn is to pick a dataset, analyze the data, and build models. I used to get stuck in tutorial hell, hopping from one tutorial to another without actually building anything. It felt like I was learning a lot, but when it came time to build, I struggled. So, I recommend finding a YouTube tutorial where the creator builds something, and follow along. Even if it feels like you don’t understand, stick with it. It will eventually start making sense after you finish a few projects and start connecting the dots. If you are a complete beginner and have no idea about Data Science or Machine Learning, build your foundation with Andrew Ng’s courses, or if you prefer reading, “Hands-on Machine Learning with Scikit-Learn, Keras, and TensorFlow” is a great book. These resources will cover the foundation you need to start building.

How do you switch off?

When life gets too hard, I play tennis but sometimes it rains in London(occasionally) . If that happens I will play an open-world game, not a gamer but I want to get lost in a rich open world ( Assassin’s Creed Odyssey) and immerse myself and forget my troubles. I also pick a painting tutorial and try to copy it (Very therapeutic for me).

What advice would you give your younger self?

I would tell my younger self not to compare with people on LinkedIn or any other social media. I would advise myself to be kinder and stay strong on bad days, reassuring that things and success will come eventually. Another important thing I would tell myself is that perfection is a myth. I would encourage myself to get out of my comfort zone and make mistakes, as nobody tends to care that much about the mistakes made.

What is next for you?

I want to work with people who are solving core AI problems and focusing on long-term results rather than short-term business impact. I’m particularly interested in multi-modal models, so I want to be in a role where I can work on these types of problems.

If you could do anything now, what would you do? Why?

Professionally, I would love to be involved in as many AI projects as possible to learn what works and what doesn’t. I also want to collaborate with organizations like DataKind, where they use data science skills to solve socio-economic issues. In my personal life, I want to start playing in local tennis leagues.

What are your top 5 predictions in tech for the next 5 years?

It’s hard to predict the future. But LLMs are all the rage now and for a good reason, it is the most knowable entity that ever existed which was created by humans.

So LLMs are here to stay, and they will get even better, with fine-tuning and other techniques, it will get human-level performance at doing specific repetitive tasks.

There will be more collaboration between robotics development and foundation models.

Drug discovery methods will change, and more AI models will be used as research tools to discover new drugs to cure diseases.

Extended Reality (XR) — AR and VR are getting better, with apple vision pro and Meta working on putting it in the hands of everyday consumers there will be a new field of making software and apps for these devices.

There is one certain thing that will happen: the cost of inference of the generative model will come down significantly and Foundation models will become much cheaper to run in production.

Another specific startup to look out for is Perflixty Ai which is disrupting how we search things online. Not sure if they will replace Google but they are getting good at showing results based on the context of the questions.

Watch Arjun’s session at the Data Science Festival here.

Thank you to all our wonderful speakers for taking part in our Speaker Spotlight!

Want to become a DSF Speaker? Apply here!