Navid Emad
Lead Rails Developer.
Agentic AI Engineer.
I ship production Rails systems and build AI agents that survive contact with real infrastructure. 12 years in, still writing code every day.
Hey, I’m Navid
12 years building and running business-critical Rails apps: payments, hardware integrations, background job fleets, and real users who notice when things break.
Today I split my time between lead engineering at Yespark and wiring AI agents (Claude, MCP, GitHub automation) into production workflows, with the reliability and cost constraints a demo never faces.
I contribute to open source because it’s the best code review I know: my PRs get judged by Rails reviewers, Datadog engineers and browser-engine maintainers, not by a portfolio page.
Specialties
- Rails at scale: queries, indexes, background job architecture
- Agentic AI in production: Claude, MCP, skills, GitHub automation
- Frontend that ships: Hotwire, Tailwind, server-rendered by default
- Observability: Datadog APM & profiling, upstream contributor
How I work
- PRs reviewers want to read: small, argued, benchmarked
- I say no: to unprofitable AI, to features that shouldn’t exist
- Frank by default: wrong fit? I’ll tell you on the first call
Numbers you can verify
Every number below is one click away from the raw GitHub search that produces it.
Where the code landed
Lightpanda
Web platform internals: CDP protocol events, WebAPI surface (DataTransfer, focus navigation, readystatechange), relative :has() selectors, WebSocket Origin handling, HTTP redirect edge cases.
see the merged PRs →Rails ecosystem
CLI generator fixes, nil-logger support in Solid Cable, an assets:precompile build-skip option, CSS build wired into spec:prepare.
see the merged PRs →capybara-lightpanda
Capybara driver to run Rails system tests on Lightpanda, the headless browser engine I contribute to. It bridges the two ecosystems above.
Datadog dd-trace-rb
Two production-crash fixes merged upstream: a heap-profiling SIGSEGV on Ruby 4, and distributed-tracing header injection broken by http.rb 6.x.
see the merged PRs →Covidliste
Volunteer work during the pandemic: landing pages for health professionals, transactional email overhaul, payment status handling.
see the merged PRs →Yespark
Parking rental at France scale: query and index optimization campaigns, Sidekiq fan-out architectures, ANPR camera fleet integrations, AI-assisted back-office tooling.
Case study · a Datadog-driven query & index campaign
- 52 s → < 1 ms for the worst production query: one composite index on a 5.6 GB payments table, measured with EXPLAIN ANALYZE in prod
- ~100 queries → 2 on an invoice listing: an ActiveStorage N+1, preloaded away
- p95 6-16 s → < 3 s across admin & API endpoints: partial indexes, cached aggregates, and OR-of-subqueries rewritten into index-backed BitmapOr plans (conservative estimate)
How I can help
I take on a small number of freelance engagements alongside my lead role.
- Rails performance & architectureaudits, query optimization, job pipelines
- AI agents in productionfrom use-case audit to agents wired into your tools
- Fractional leadtechnical direction, review culture, mentoring
Tell me what you’re building, a few short lines is enough. I reply within 24 hours, frankly: what I can do and how I’d approach it. If I’m not the right person, I’ll say so and point you elsewhere if I can.