Justin Thomas

AI-Driven Architecture & Engineering Leadership

Ex-AWS · Stanford AI/NLP certified · Fortune-100 engineering leader. Two decades building and leading backend platforms, SRE, and edge — now shipping generative AI, resilient APIs, and agents.

Experience

Ford Credit
Principal Architect
May 2024 – Present · United States
  • Lead architecture for Ford Credit's emerging AI space — driving 90% adoption of AI/agentic coding across engineering.
  • Shipped a reusable RAG call-center platform in 2 months, integrating knowledge bases to streamline support workflows.
  • Lead AI transformation org-wide — routinely presenting to 30–70 engineers in the cross-functional AI Tech Guild and Architecture Forum.
  • Shipped pilot AI agent flows integrating Jira and other sources to automate updates and ticket compliance.
  • Helped hire 4 AI engineers and set software-excellence standards: CI/CD, linting, and structured AI/ML hiring panels.
Amazon Web Services (AWS)
Software Engineering Manager
Oct 2022 – Apr 2024
  • Led 5 engineers delivering third-party integrations for CodeCatalyst & CodeSuite.
  • Drove region expansion across 5 global regions, aligning 30 services with 16 teams for on-time re:Invent launches.
  • Built a Kotlin ticket-ingestion system; ran security reviews; mentored and promoted 2 engineers.
Verizon Media (→ Yahoo / Edgio)
Director of Engineering
Oct 2018 – Sep 2022 · Los Angeles, CA
  • Built and scaled SRE, platform & EdgeCompute orgs from 2 to 22 engineers, growing internal leaders.
  • Migrated 500 engineers to Kubernetes + GitLab CI/CD — Time-to-Ship from 3 months to 1 day.
  • Launched "Functions at Edge," a globally available serverless edge product.
  • Implemented SLOs that cut time-to-detection from 15 min to 2 min.
Red Bull Media House
Director of Engineering → Lead Software Engineer
Jan 2013 – Oct 2018 · 5 yrs 10 mos · Santa Monica, CA
  • Designed an nginx API gateway for Red Bull TV's microservices — API latency 700ms → 100ms.
  • Migrated the platform from Java to Golang: 8 → 1,000 RPS (125× throughput), operational cost −90%.
  • Moved infra from Heroku/AWS to GKE; adopted Pub/Sub, CloudSQL & BigQuery over Kafka, cutting DevOps cost 80%.
  • Stood up Grafana & Kibana monitoring (−50% detection time); data-mined artist social signals to sign 10 emerging artists.

Earlier

Chromatik — Lead Software Engineer
Front- & back-end for a sheet-music reader + recording/sharing platform (Backbone/canvas, Node.js, RabbitMQ, MongoDB).
2011–2013
Rally Software — Sr. Software Engineer, AgileZen
Shipped Rally Scrum-tool features (Java, Oracle, Spring) and rewrote the AgileZen backend + its Node.js/RabbitMQ notification service.
2010–2011
Verizon Business — Sr. Software Engineer
Redesigned network-management architecture; Lucene search engine 20s → 10ms; data load 7 min → 5 s; automated ticketing.
2006–2010
bioMérieux — Software Engineer
Ruby on Rails & Spring web apps generating pathogen growth curves.
2006
Honeywell Aerospace — Software Intern
Built "six-pack" test-bench visualization software for CFM turbine engine controls on the fuel-control test line.
2000–2001

Selected Work — westcot.io

Hermes Observability — open-source AI-agent ops stack
Prometheus/Grafana cost, latency & error monitoring for LLM-agent sessions — per-session cost metrics, alerting & a deadman heartbeat. Extracted into a reusable repo.
github
Agent Deck Bake-off — multi-LLM agent pipelines
Agents that generate complete slide decks; benchmarked four frontier models on one brief for cost and quality.
LLMs
Quadcopter Drop Bay — servo-actuated mechatronics
Designed, 3D-printed & programmed a servo-released payload bay for an autonomous drone — CAD-modeled, with an interactive three.js kinematics viewer.
3D · servo
⬇ Download PDF ← Back to site