I am interested in model post-training, efficient AI research, and the practical systems that make smaller language models more capable.

Currently, I am doing my RL internship at Wootzapp, where I create tasks, build verifiers, and help ship RL environments.

I like hacking around problems worth solving: reinforcement learning, evaluation, tooling, and ways to make AI systems more useful without making them unnecessarily large.

I would be more than happy to chat about the frontier, discuss ideas, collaborate on problem statements, and work on interesting problems. My email is ishagarg.1103@gmail.com.