I design, modernize, and operate software that sits directly in the path of real business traffic.
My work includes search infrastructure serving hundreds of e-commerce sites, high-volume automotive fitment APIs, ingestion pipelines processing millions of product records, large relational databases, and AI tooling connected to production systems.
I am most effective where domain complexity, legacy architecture, and reliability requirements collide. I care about measurable performance, observability, safe migrations, strong testing, and systems that remain maintainable long after the initial implementation.
Designing and operating multi-tenant search infrastructure for large e-commerce platforms.
My work spans index architecture, relevance, faceting, filtering, caching, load testing, zero-downtime migrations, and production operations. I have worked with both managed and self-hosted search platforms, including infrastructure delivering sub-15ms hot p95 latency.
Working deeply with the ACES and PIES ecosystems, including VCDB, PCdb, PAdb, Qdb, vehicle fitment, taxonomy, product normalization, and validation.
Aftermarket data has its own class of difficult engineering problems: incomplete standards coverage, inconsistent supplier data, millions of product-to-vehicle relationships, and systems that must remain accurate across constantly changing catalogs.
Building MCP servers, agent workflows, and internal AI platforms that connect language models to source control, business data, and operational systems.
The interesting work is not calling a model API. It is building the surrounding infrastructure: authentication, authorization, policy enforcement, context management, observability, evaluation, failure handling, and cost control.
Modernizing legacy applications, designing APIs and background workers, tuning databases under production load, and improving the reliability of business-critical systems.
This includes schema design, query optimization, queue architecture, incident response, telemetry, automated testing, deployment safety, and incremental migrations that cannot interrupt existing users.
| Project | Description |
|---|---|
| Openleaf | An open-source, ProseMirror-based editor designed for integration into existing content platforms, with structured APIs and agent-accessible tooling. |
| ACES 4.2 Validator | Open-source validation tooling for ACES XML, built with Rust and DuckDB, with SARIF output for CI integration. |
| Nebula | A statically typed language for authoring AI agents that compiles to Python. The compiler is written in Rust. |
Languages: TypeScript, Go, Python, PHP, Rust, SQL
Backend: Node.js, FastAPI, FastMCP, CodeIgniter
Data: MariaDB, MySQL, Redis, DuckDB
Search: Meilisearch, Algolia
Infrastructure: GCP, Docker, Linux, Nginx, Plesk, GitHub Actions
Domains: E-commerce, automotive fitment, search, data pipelines, AI infrastructure
I prefer production evidence over assumptions.
That means profiling before optimizing, designing migrations with rollback paths, treating observability as part of the feature, automating repetitive verification, and using AI to accelerate engineering without outsourcing engineering judgment.
I am always interested in conversations around search infrastructure, automotive aftermarket data, platform modernization, AI tooling, and difficult production systems.