🇨🇦Tanmay
Bakshi

Tanmay Bakshi was five years old when he wrote his first computer program. His parents, both software engineers who had emigrated from India to Toronto, did not push him toward coding — it was simply in the air of the household, a language spoken by people he admired, and he decided, with the matter-of-fact logic of a five-year-old, that he would learn it too. By nine, he had built and published his first iPhone application. By twelve, he had done something that made IBM Watson's engineering team do a double-take: he had found a bug in one of the most sophisticated artificial intelligence systems on Earth, and he had built on top of it.

IBM Watson, named after the company's founder, was at the time the most famous AI system in the world — the machine that had defeated Jeopardy! champions Ken Jennings and Brad Rutter in 2011 in a televised contest that shocked the public with a demonstration of natural language question-answering capability. Watson's underlying architecture — a system that could read millions of documents, extract relationships between concepts, and weigh evidence to generate confident answers — represented the state of the art in enterprise AI. Tanmay Bakshi, eleven years old, watched a documentary about it and decided to build an application on top of it within a week.

The application he built, "Ask Tanmay," was designed as a web-based natural language question-answering system: you typed a question, and it returned the best answer Watson could generate, ranked by confidence. Getting Watson to work required navigating IBM's developer APIs — interfaces designed for professional programmers, with documentation written for adults working in enterprise technology environments. Bakshi navigated them, built the application, and in the process identified a bug in Watson's behavior. He reported it to IBM. IBM's engineers, apparently impressed by the source of the bug report, invited him in. He became, at twelve, the world's youngest IBM Watson developer.

IBM named Bakshi an Honorary Cloud Advisor. He was given access to Watson's development environment and worked on applications that ranged from educational tools to healthcare. His YouTube channel, "Tanmay Teaches," accumulated hundreds of thousands of subscribers as he produced tutorial videos on machine learning, Swift programming, IBM Watson APIs, and deep learning — explaining concepts that most educational content reserved for university students to an audience that included teenagers, self-taught programmers, and curious professionals. The Toronto Star named him its Newsmaker of the Year in 2013 when he was still nine years old. By fourteen, he had published a book: Hello Swift! iOS Programming for Kids and Other Beginners.

In 2017, he gave a TED talk titled "Technology that Tackles the Teen Suicide Epidemic." The project behind it was called Cognibuddy — an AI system that monitored online behavior patterns to detect the warning signs of suicidal ideation in teenagers, building an early intervention tool for a crisis that was claiming thousands of young lives annually. The talk was not a theoretical exploration; it described a functional prototype that Bakshi and his team had built, a system designed to flag at-risk individuals to counselors and parents before a crisis point was reached. He was fourteen.

Bakshi continued developing his technical expertise through his teenage years, moving from IBM Watson to a broader engagement with deep learning, neural network architectures, and the mathematical foundations of modern AI. He became a Google Developer Expert, spoke at conferences in New Zealand, Finland, Denmark, and Australia, and collaborated with research teams studying the intersection of AI and healthcare. By his early twenties, he had joined research circles connected to Google and DeepMind, working on the frontier of large model research that was rapidly reshaping the entire technology landscape. The child who decoded IBM's API documentation at eleven was now contributing to the systems that would define the next decade of artificial intelligence.