AI engineer & systems architect · RAG + stateful orchestration

AI systems, built to hold up.

AI engineer and systems architect building governed, observable, and reliable AI products.

What I do

I work where model behavior meets the systems that govern, recover, and explain it.

AI produces candidates and handles ambiguity only inside bounded workflows.
I decide system boundaries, evidence authority, recovery behavior, and human approval points.
I verify failure paths and evidence before treating system behavior as reliable.

Stack

My focus is system architecture and AI orchestration: stateful agent workflows, retrieval, and the durable, multi-tenant services that run them.

System architecture: Multi-tenant isolation, versioned backend contracts, idempotent retries, webhook reconciliation, and replay snapshots
AI orchestration: LangGraph, Temporal, BullMQ, MCP, multi-provider LLM workflows, and worker orchestration
AI and retrieval: RAG, LLM evaluation and scoring, prompt engineering, prompt and citation provenance, and the OpenAI and Anthropic APIs
Languages: TypeScript, Python, JavaScript, Java, and SQL
Backend and web: NestJS, Node.js and Express, Spring Boot, Next.js, React, React Native, and REST APIs
Cloud and data: AWS ECS and Fargate, Lambda, S3, RDS, and Cognito, with PostgreSQL, MySQL, MongoDB, Redis, Prisma, and Docker
Security and testing: OAuth2, JWT, RBAC, SSRF controls, secret redaction, Jest, Playwright, and CI gates

Selected work

Selected AI systems, from my personal products to my contributions on team-owned platforms.

Team contribution
Governed AI Orchestration
I designed and implemented stateful orchestration contributions for governed AI workflows, separating resumable execution state from evidence authority.
Personal project
Companion Studio
I built a local-first desktop AI runtime that packages its own inference stack.
Team contribution
AI Platform Reliability Contributions
I contributed reliability improvements to a multi-tenant AI platform, focusing on scheduler recovery, atomic evidence writes, cancellation-aware APIs, request-budget correctness, and workflow efficiency.

01
Personal project
Reusable AI Workflows
Two public repositories turn report writing and code-edit mentoring into explicit, reusable AI workflows.
- Open the public report-writing workflow repository
- Open the public code-edit mentoring workflow repository
Team contribution
Multi-provider Compliance Platform
My contribution covered webhook ingestion, workflow gates, administrative workspaces, and audited document delivery.
AI-generated material remained evidence, while deterministic policy produced workflow decisions.
Personal project
Local Travel Planner
A local travel-planning MVP uses deterministic parsing for ordinary plans and an on-device model only when input is ambiguous.
Strict event validation runs before any plan is persisted.
04
Team contribution
Open-source Dash Components
I contributed a reusable component library to an open-source Dash ecosystem project during an internship.
- Open the public Dash component library repository

Experience