Alex Lewis
Senior Software Engineer, Zapier
https://alexmlewis.com | https://github.com/mpaarating | https://www.linkedin.com/in/alex-lewis-2157759b/ | alexmichael.lewis@gmail.com

SUMMARY
Senior software engineer with 10+ years shipping production software across frontend, backend, platform, and automation. Current focus: applied AI — agent infrastructure, LLM evaluation, retrieval, tool use, and developer tooling — building the systems that make LLM-powered products dependable.

SKILLS
- Applied AI: AI agents & orchestration, LLM evaluation (LLM-as-judge, ground-truth measurement), Retrieval-Augmented Generation, Tool calling & MCP, Prompt architecture, AI observability
- Languages: TypeScript, Python, JavaScript, SQL
- Frameworks & Platforms: React, Next.js, Node.js, PostgreSQL, AWS, Vercel
- Practices: CI/CD, API design, Software architecture, Production observability, Serverless

EXPERIENCE
Senior Software Engineer, Zapier (Aug 2025 - Present)
  - Built an internal agentic end-to-end testing platform — agent runtime, CI integration, and a plugin system that decoupled it from any single app so agentic tests could run against many applications; benchmarked the agent at ~90% first-run pass rate across 50 runs.
  - Designed evaluation infrastructure for an LLM-as-judge system: deterministic verdict guards plus a ground-truth measurement layer that cross-tabs eval verdicts against real merge outcomes to correct judge bias and verify verdicts track reality.
  - Built an agent-orchestration harness with live permission prompts, an egress sandbox, self-supervision, and cost metering; cleaned it up and handed it off as a shared, contributable team repo.
  - Built a semantic-release CI/npm pipeline so any repo could consume the AI testing agent via npx (adopted across 4 repos), and made agent session transcripts first-class test-proof artifacts for observability.
  - Shipped inline app authentication to all non-enterprise users, removing auth friction from an AI assistant experience; built and evangelized a Claude Code plugin catalog for the org (demoed to 20+ engineers).
  - Cross-team infrastructure fixes: a one-line change unblocking Sentry setup for every Next.js service in the internal deploy pipeline, and diagnosing a cross-service auth failure (Vercel Secure Compute + PrivateLink).

Senior Software Engineer, Your Money Line (Oct 2023 - Jun 2025)
  - Built a production Retrieval-Augmented Generation (RAG) system for personal-finance guidance, with chunking and prompt-engineering focused on accuracy and consistency on sensitive topics.
  - Ran a lightweight grading/evaluation signal that informed which prompts the system used in production.
  - Prototyped an LLM question-answering assistant (serverless) and built a generative-AI product feature.
  - Built an AWS Textract document-AI pipeline for loan-payment processing, and a text-to-speech pipeline.
  - Designed a serverless HRIS data-sync integration (AWS Lambda, API Gateway, Step Functions, SQS FIFO with idempotent dedupe, Aurora) supporting bulk import and near-real-time webhook delta sync.

Software Consultant, Independent (2020 - 2021)

TARGET ROLES
Senior Applied AI Engineer, Staff Software Engineer, AI, Senior AI Product Engineer, AI Platform Engineer, Agent Infrastructure Engineer, Senior Full-Stack AI Engineer