WORK / AREAS OF PRACTICE

Building systems that
make it to production.

A look at the technical areas I work in, from retrieval and ranking to reliable AI agents. These are high-level summaries, not disclosures of proprietary implementation details.

4 areas shown

01 / ML Systems

Recommendation systems

Retrieval, ranking, embeddings, and building fast personalization pipelines.

I work on large-scale recommendation and advertising ranking systems. This section discusses public engineering concepts rather than implementation details of any employer’s internal systems.

02 / Agentic AI

Reliable AI agents

Orchestration, tool use, retries, evaluations, and systems that recover gracefully.

My focus includes production-minded tool use, state management, evaluation, and safety in multi-step agent workflows. Specific internal implementations are intentionally not disclosed.

03 / ML Infrastructure

Model training & serving

Scaling training, inference, and monitoring for practical LLM applications.

Areas of experience include distributed PyTorch training, model serving, and cloud infrastructure. Public writing here describes the technologies and approaches, not proprietary architectures.

04 / Applied Research

Multimodal & generative AI

Connecting vision, language, and generative techniques to useful applications.

My past applied research spans NLP, generative AI, computer vision, and reinforcement learning. This portfolio shares high-level areas rather than confidential company material.

TECHNICAL WRITING

Ideas, tested
in public.

Deep dives on AI agents, retrieval pipelines, and applied ML.

Browse notes ↗

Building systems thatmake it to production.