About

I’m Yasaswi, a data scientist at Neuralix.AI. I work on large language models: model architecture, distributed pretraining, and post-training with supervised fine-tuning, preference optimisation and reinforcement learning. I also make our models work with Hugging Face and vLLM.

I’m most interested in how architecture choices affect training and inference, and in using reinforcement learning to improve how models behave on real tasks.

Projects

In my own time, I post-trained Qwen2.5-1.5B with reinforcement learning (DAPO in verl). I trained two teachers, one for grounded RAG answering and JSON output and one for math and instruction following. I then merged them into one student with multi-teacher on-policy distillation. The student beats the original model on RAG, instruction following, JSON extraction and math, and isn’t significantly worse on any of the 24 benchmarks I tracked. The code and write-up are on GitHub, and the models and training data are on Hugging Face.

Background

I studied chemical engineering at JNTU Hyderabad and, alongside it, completed a BS in Data Science and Applications at IIT Madras. I then spent almost two years at TCS leading a support team for an enterprise CRM, and joined Neuralix.AI in 2026.

Outside work

I enjoy driving, follow tennis, football, basketball and F1, and read manga and light novels. I also have a dog, Kali.

Contact

Email is best: yasaswi.kapila@gmail.com. I’m happy to talk about LLM training and reinforcement learning. You can also find me on GitHub and Hugging Face.