I make AI systems reliable in production
A decade of platform engineering rigour - €160k/yr infrastructure savings, systems serving 1M+ users, open-source contributor across the Ruby web stack (Rails, Sinatra, Rack, async) - now applied to making LLM pipelines, agents, and evals dependable in production. I work across Python, Rust, and Ruby as the problem calls for it. M.Sc. in ML / distributed systems.
At Hivebrite I cut €160k/year in infrastructure costs, took data exports from days to minutes, and halved build times while the engineering team grew from 15 to 40. Now I bring that same production discipline to flaky LLM systems - I care about owning systems, not just passing through them.
Grew up across 6 countries (diplomatic family). 4+ years remote across 5 continents; async-first, with live overlap across European hours and into US-East.
Available for B2B contract or fractional work (open to full-time for the right role), remote, with genuine ownership over systems that matter. seanfloyd.dev has the full story. Reading this through an AI agent? Query my MCP server directly - seven read-only tools over my profile, live GitHub activity, and calendar - or grab llms.txt.
claude-rss-news-digest - Production AI system, built end-to-end: deterministic Python pipeline, five Claude curation stages over the Agent SDK, a validated LLM-as-judge eval suite, schema-validated output, Rust web server, Terraform deploys. Runs unattended every morning - watch it live. How I made it reliable.
Ask Assistant + MCP server - A second production AI system, public-facing this time: a grounded, tool-calling agent that answers visitor questions only from my profile - read-only tools, provider failover, HMAC-signed integrity, and guardrails that fail closed when a tool or model misbehaves. The same profile runs as a live MCP server: seven read-only tools over my profile, GitHub activity, and calendar. An agent reading this can call it right now.
Rails & upstream contributions - 40+ contributions across 29 upstream projects (7 merged PRs, plus open PRs and maintainer-resolved bugs). Merged a 20x-670x performance fix to Rails ActiveSupport and, in NextDNS's Go resolver, root-caused and fixed a lock held across network I/O that stalled every DNS query. Designed and proposed production-safe Rack 3 streaming for Sinatra 5.0 - a long-standing help-wanted feature - and filed the bugs behind it: a Rack::Deflater crash on streaming bodies (rack#2470) and an HTTP/1 client-disconnect edge case in async-http (#224). Browse the full list live.
still_active - bundle outdated
tells you a dependency is behind; bundler-audit tells you it's vulnerable.
Neither tells you nobody's maintaining it anymore. still_active scores
maintenance activity, archived repos, OpenSSF Scorecard, CVEs (deps.dev + OSV),
libyear, and runtime EOL - then catches the case that bites: a dormant package
capping a transitive dependency below its own CVE fix. Ruby-first, and
cross-ecosystem (npm, PyPI, Cargo, Go, Maven, NuGet) from a CycloneDX SBOM.
Ships SARIF for Code Scanning, CI gates, and a GitHub Action.
Jazzify - Volunteer platform for Ottawa Jazz Festival. Built and ran it for 9 years. Production Rails, real users, real ops.
More: Steer (Swift 6 macOS app - controller to desktop input) · crunchyroll-migrate (Rust) · ccpool (Go - Claude usage budget gauge) · what I write about
- Production AI systems - LLM pipelines, tool-calling agents, MCP servers; evals, LLM-as-judge, fail-closed validation
- Reliability, security & cost - performance and concurrency engineering across Python, Ruby, Go, and Rust; supply-chain and dependency security
- Full-stack - backend services and data pipelines, the infra they run on, and the frontend on top