India needs a stronger olympiad pipeline

Modal Labs co-founder and CTO Akshat Bubna, who first came into the spotlight as the first Indian to win gold at the International Olympiad in Informatics, says the problem-solving mindset he developed as a teenager continues to shape how he builds infrastructure for the next wave of AI companies.

India, he believes, has long undervalued global Olympiads. “The community is small, but many of the people who compete go on to build remarkable companies,” Bubna said in an interaction with us, pointing to the founders of Cognition, a $10-billion AI firm, as fellow IOI alumni. Several of Modal’s hires also come from competitive programming backgrounds.

“Programming competitions teach you that most problems are solvable,” he said. “You also learn to work incrementally – start with something simple that works, then improve it over time. That’s exactly how we’ve built Modal.”

Founded in 2021 by Bubna and Swedish engineer Erik Bernhardsson, Modal Labs builds what it calls AI-native infrastructure – a software layer that allows developers to run machine-learning workloads at scale without relying on legacy systems. The company has developed its own file system, container runtime and scheduler, enabling sub-second container start-ups and low-latency routing.

In September, the New York City–headquartered venture raised $87 million in a Series B round led by Lux Capital, valuing the company at about $1.1 billion.

“Before Modal, I worked at Scale AI and saw the same infrastructure problems over and over again,” Bubna said. “Tools like Kubernetes weren’t designed for large-scale ML workloads or global GPU management. People keep trying to retrofit old systems for new use cases, but building from scratch lets you design for today’s needs.”

Modal aggregates GPU and CPU capacity across cloud providers including AWS, Azure, Oracle and Google Cloud, allowing developers to deploy code instantly without managing servers. “Our focus is developer experience,” Bubna said. “With Mod-al, you can make a change and deploy it globally in seconds. That speed really matters to teams.”

The company’s customers include Meta, Substack, Suno, Scale AI and India’s Zomato. Its pay-asyou-go model allows enterprises to scale thousands of GPUs for short bursts, aligning costs closely with actual usage. Modal’s platform spans five core products – Inference, Training, Batch, Sandboxes and Notebooks – covering the full machine-learning lifecycle.

Bubna said GPU availability remains a near-term constraint for the AI industry, though he expects that to ease within the next two years. “The bigger bottleneck is talent,” he said. “Despite all the talk about AI replacing engineers, you still need very strong people to build these systems.”

Modal, which now has around 70 employees across New York, San Francisco and Stockholm, continues to recruit heavily from programming-competition networks. “We hire deliberately from these communities because they tend to have a high ceiling in software design and problem-solving,” Bubna said.

Reflecting on India’s ecosystem, he said limited research infrastructure makes it difficult to build deep-tech companies locally. “When I wanted to study machine learning, India didn’t feel like the place where you could do that,” he said. “People do well at IITs, then move elsewhere to pursue research. Fixing that gap is crucial.”

Asked what advice he would give a 16-year-old Olympiad aspirant today, Bubna was direct: “It starts with believing you can do it. There’s so much talent in India – if we push harder in this direction, we can do far better.”