Founding ML Engineer (Computer Vision) - Gurgaon - Onsite (Mon-Sat)
What You'll Own:
- The full ML lifecycle: data engine, training, evaluation, deployment, monitoring, and the loop tying them together
- A serious data engine: collection, labeling strategy, hard-example mining, and test sets that predict real-world performance
- Computer vision that must be right in real time in the hardest conditions (bad lighting, blocked views, uncontrolled angles, hundreds of subtly different processes)
- Full authority over how ML is done at a company where ML is the product, working directly with the founders
Requirements:
(A) Must Have -
- Trained computer vision models yourself (own training runs, own data decisions, not fine-tuning via API) and deployed them to production
- Built or run a serious data engine: collection, labeling strategy, hard-example mining, evaluation sets tied to real-world performance
- Owned an ML system end to end in production (data, training, evals, deployment, monitoring), iterated against real-world failure, not benchmark leaderboards
- Deep computer vision fundamentals: can reason from first principles about why a model fails on real video and design the fix
- Research depth with proof of shipping (publications/citations are great, but only paired with production; a pure academic is a no)
- Did your best work at an early seat in a high-performing, US-based startup: small team, no safety net, outcomes on you
- Comfortable with 6 days a week, in office, in Gurgaon
(B) Nice to Have -
- Published research with real impact
- A system people said couldn't be built, or a result that made a previous team's product possible