AI Engineer at Otsuka Corporation in Tokyo. I spend most of my time on large language models, agentic systems, and retrieval-augmented generation for real enterprise workloads — the kind that need to hold up under production traffic, not just in a demo.

Day to day that means fine-tuning models with SFT and RLHF, serving inference with vLLM on internal GPU clusters, and building RAG pipelines where latency, access control, and evaluation rigor actually matter. I probably care more about the unglamorous plumbing than the model itself.

Outside of work I open-source tools I wish had existed when I needed them: a multi-agent chat platform with persistent memory, a YAML-driven RAG benchmark harness, and a dataset augmentation pipeline for SFT and preference data at scale. I studied Computer Science at IIT Patna, and my research on transformer models for disaster-event extraction was published in Online Social Networks and Media.

Before any of this, I did a lot of competitive programming. I think that’s why I still enjoy debugging something down to the exact line instead of accepting “it works on my machine.”

Languages & frontend

ML / AI

Databases & search

MLOps & infra

• Agent Forge — multi-agent chat platform with persistent memory and tool calling.
• RAG Benchmark Toolkit — compare chunking strategies, retrievers, and rerankers across hundreds of configs instead of guessing.
• Dataset Augmenter — generate SFT and preference data at scale.

I’m always happy to chat about LLM fine-tuning, RAG architecture, or agent design.

• Website: rahilop.github.io/profile
• Email: rahilrizvi0786110@gmail.com
• LinkedIn: linkedin.com/in/syedrahil876