What I’ve worked on

• LLM Research: Improved LLaMA-2-13B truthfulness to 87.1% on SycophancyEval by localizing sycophancy-related behavior through path patching and applying supervised pinpoint tuning to targeted model components.
• AI Research @ Allen Institute for AI: Built literature-retrieval and agentic research workflows using query expansion, hybrid reranking, context engineering, and Search-R1, moving LLM priors closer to empirical posteriors in 66% of trials.
• Cloud & Data Engineering: Led data-pipeline and cloud-infrastructure optimization during an AWS-to-GCP migration, reducing end-to-end pipeline latency by 30%, while building SQL, BigQuery, and FastAPI monitoring workflows that contributed to 20% higher user engagement.
• Production ML: Developed ML systems for logistics, including return-risk and address-verification models achieving 98% F1 and 87% precision, helping reduce shipment returns by 5% and out-of-delivery-area cases by 40%.

What I work with

Languages & ML: Python · C++ · Java · SQL · PyTorch · JAX AI: RAG · LLM Evaluation · Fine-tuning · Multi-Agent Systems · Personalized LLM Systems Systems: FastAPI · GCP · AWS · Data Pipelines · Distributed Systems

I’ve also built production prediction APIs and published research on LLM-powered fact checking in IEEE Access.

What I’m interested in

I’m particularly interested in Software Engineering, Data Science, and AI/ML Research Engineering roles where I can combine strong engineering fundamentals with machine learning to build systems that are technically challenging, measurable, and useful in the real world.

Always happy to connect with people working on interesting problems across software, data, and AI.

Programming Languages

Frontend Development

Backend Development

AI & ML

Databases

Devops

Backend as a Service(BaaS)