Shantanu
Tripathi

Three years into computer vision and I've learned the machines are better at spotting faces than I am at spotting deadlines.

I build systems that detect, recognize, and predict in real time — the kind of work where "close enough" gets you fired and "99%" gets you a good night's sleep.

I've done this at Metafusion, Awiros, and Bobble AI — three very different rooms, one recurring theme: teach a model to notice something a human would've missed, then ship it before anyone notices you. Vision, NLP, the occasional robot. All of it, eventually, comes down to the same question — what's actually in this frame, and what do we do about it.

Before any of that, I was President of IRIS, our robotics club, which mostly meant convincing a room of very smart people to build things that balanced, followed lines, and didn't catch fire. Leadership, it turns out, is just debugging with more people involved.

Off the clock, I play percussion, write poetry, and draw characters that will never appear in a dataset. Call it my analog fallback — for when the model's confident and I'm not.