I build agentic AI systems, RAG infrastructure, and applied LLM research projects across Python, PyTorch, FastAPI, MCP, and Next.js.
Featured
Featured
Open-source terminal AI coding-agent harness for studying and building reliable agent loops, schema-first tools, MCP integrations, skills, subagents, persistence, checkpoints, safety boundaries, and a Rich terminal UI.
A production-ready RAG-powered document intelligence platform that enables users to upload documents, chat with AI using intelligent retrieval, and generate content with proper citations. Features hybrid search, reranking, and secure document processing.
Research
A compact proof line for the research behind the systems.
Featured paper · arXiv preprint · 2026
Geometry-aware instruction tuning
Mohit Goyal, et al.
Paper-to-code
PyTorch implementation of the LLaMA architecture, kept as a compact proof that I can translate model papers into working code.
$ whoami
ML Engineer / Agentic AI Builder / Researcher
I build agentic AI systems, RAG infrastructure, and applied LLM research projects that move from paper-level ideas into usable products.
Most of my work lives at the edge of research and shipping: reliable systems, not just impressive demo windows.
proof: 3 publications / 15x national hackathon winner / research-to-prod AI systems
focus:
stack:
github:
@MohitGoyal09 github activity
Stripe's agentic commerce stack is moving payment authority from seller-controlled checkout pages to buyer-controlled wallets, changing how online commerce works.
Transformer-Squared enhances AI with dynamic adaptation, boosting performance across tasks and reducing computational costs
Explore Titans, the next-gen AI memory architecture, revolutionizing long-term context and task efficiency by mimicking human memory systems
Hey, you scrolled this far, let's talk.
“A man who is master of patience is master of everything else.”
~ George Savile
Made with ❤️ by Mohit Goyal
© 2026. All rights reserved.
