Skip to main content
Now: Building the AI Code Reviewer · Production

I build AI systems that survive production.

Senior Full-Stack Engineer specializing in AI integration, robust platforms, and deterministic outputs from stochastic models. Stack: React, Next.js, Node, Go, Python.

5

Years Building

10+

Products Shipped

4+

AI Integrations

Selected Work

Flagship Systems

AI Code Reviewer

Production
ShippedMeasured

Build Status

All build phases complete and tested (426 tests in CI). GitHub delivery (packages/github), the acr-review CLI, and the self-review GitHub Actions workflow are merged to main — the agent reviews this repository's own PRs in CI (19 workflow runs as of 2026-07-11).

An autonomous agent that reviews GitHub pull requests with code-aware retrieval and tool use

TypeScriptNext.jsPython+9
View case study
View AI Code Reviewer

Multi-Agent Customer Ops Platform

Production
100%Measured

Traceability

Achieved full request-level tracing for all agent invocations, enabling operators to audit any automated decision.

An autonomous multi-agent orchestration layer that resolves complex customer operations tickets.

TypeScriptNext.jsXState+3
View case study
View Multi-Agent Customer Ops Platform

Telecom POS Platform

Production
HundredsScope Fact

Terminals

Successfully deployed and actively powering hundreds of retail terminals across North America.

A unified, multi-tenant Point of Sale system powering hundreds of retail terminals across North America.

React Native WebTypeScriptRedux Toolkit+2
View case study
View Telecom POS Platform
View All Systems

From the Notes

Thoughts & Insights

Architecture

Building a Portfolio as an Engineering Lab

Why I rebuilt my portfolio from scratch without a CMS, and how I turned it into a verifiable system to showcase architecture, design, and continuous integration.

3 min read
AI

Shipping the Boring Half: AI Code Reviewer, Phase 1 Foundations

What 'production-grade scaffolding' actually looks like for an AI app in 2026 — monorepo, project references, env-gated observability, and a streaming endpoint with a placeholder agent — before any retrieval or tool use lands.

11 min read
AI

Building an AI Code Reviewer That Actually Reads Your Code

How I'm building an AI agent that reviews GitHub pull requests with hybrid retrieval, tool use, and real evals — without LangChain or framework glue.

10 min read
View All Notes

Engineering Principles

Claims Backed by Systems

Retrieval-First AI

LLM quality is bounded by retrieval quality. The AI Code Reviewer uses AST-aware chunking, hybrid BM25 + vector search, and cross-encoder reranking — not prompt stuffing. An eval harness with a committed baseline measures whether each architecture change actually improves review quality.

Proven in: AI Code Reviewer →

State Machines for Non-Determinism

LLM agent outputs are non-deterministic. Standard state management produces brittle race conditions. The Multi-Agent Ops platform models agent states explicitly with XState — handling tool failures, human interrupts, and unexpected LLM events without a catch-all error boundary.

Proven in: Multi-Agent Customer Ops →

Single Codebase, Production Scale

The Telecom POS powers hundreds of retail terminals across North America — iPad and desktop — from one React Native Web codebase with 95% code sharing. A Node.js BFF translates 15+ legacy SOAP APIs into a clean GraphQL surface without a costly backend rewrite.

Proven in: Telecom POS Platform →

Let's Build Something

Open to senior full-stack and AI-engineering roles, as well as selective client projects. Reach out and let's chat.

Get in Touch