Akhil Shekkari
AI engineer working on LLM post-training, inference systems, and agents.
I build and evaluate machine-learning systems end to end, from post-training and tool-using agents to inference runtimes and computer vision.
Selected work
Mathematical ReasoningSFT and Dr. GRPO for Qwen2.5-3B
Sandboxed Coding AgentLangGraph CLI with repository-scoped E2B sandboxes
Inference Engine from ScratchPaged KV cache, block allocation, and continuous batching
Diffusion Image Super-Resolution4× image reconstruction with Swin attention
All work →
Recent writing
Lessons I Learnt from Building vLLM Internals
Context in LLMs: what determines it and what it costs
GPU Fundamentals & LLM Inference Mental Models
All writing →
Background
MS in Applied Machine Learning at the University of Maryland. Previously worked on a Pfizer engagement at Atrium and spent three years building ML and data systems at Tezo.