Elijah Lewis
New York City·hello@elijahlewis.dev·linkedin·github·elijahlewis.dev
I'm a product engineer who ships 0→1 tools for small, mission-driven teams, and I lead AI-first, agentic engineering practice (Claude Code, MCP, custom internal tooling). I work full-stack, from healthcare-interoperability legacy systems to modern serverless stacks, and I volunteer in civic tech, where my work is cited in human rights reports.
Experience
First Round Capital— top-tier seed-stage VC — Uber, Notion, Roblox
Oct 2024 – PresentSoftware Engineer
New York City- One of two engineers firm-wide, owning end-to-end design, data, infra, and vendor decisions for the firm's internal product platform; shadow partners during live deal flow meetings to inform product decisions.
- Built the firm's internal MCP server (the "company brain"), giving AI agents governed access to firm data and business context across tools, and set the team's modern engineering stack (TypeScript, AWS serverless, Postgres + Supabase).
- Co-shipped "At A Glance," the firm's daily-driver deal review tool used by every partner, chief of staff, and investor; described internally as mission-critical and credited with helping the firm review a record-high 10,000+ investment opportunities in a single year.
- Co-designed and built "First Source," scanning the open web and third-party APIs (Harmonic, Twitter, etc.) to surface emerging startups; aggregates 200+ AI-enriched signals weekly for structured human review.
- Designed and shipped a custom IDE for the engineering team: a desktop app that provisions each engineer a dedicated EC2 environment running Claude Code with specialized skills and port-forwarding to their local machine.
- Created agentic development workflows (Claude skills) that scope a Linear ticket, plan it with Codex, present the plan as an interactive HTML spec, review the code, and open a PR — used daily inside the custom IDE.
- Built two Linear AI agents — a scoping agent that explores features and presents mockups, and a builder agent that scopes, asks clarifying questions, delivers an HTML plan, and submits a PR entirely via Linear comments, validating features with Playwright MCP screenshots.
TypeScriptAWS serverlessPostgresSupabaseMCP
Inner File Software— human rights investigations
Jan 2024 – PresentHuman Rights Volunteer Software Engineer
Remote- Built a public-facing data visualization (museum exhibit, viral on Reddit/TikTok) and internal research tooling that doubled investigator throughput; work has been cited in major human rights reports supporting advocacy and international accountability.
Codestache LLC— personal umbrella for independent products and prototyping
Sep 2022 – PresentFounder
Remote- Launched Modal (usemodal.com), a productivity SaaS featured on Product Hunt, and Bookish, an AI semantic search engine over 9,000+ books using vector databases, HuggingFace, and LangChain — shipped before semantic search hit the mainstream.
Vector databasesHuggingFaceLangChain
4RL— early-stage social startup combating the loneliness epidemic
Apr 2024 – Oct 2024Founding Engineer
Washington, D.C.- Shipped the initial beta app and marketing site; drove early validation with 300+ in-person connections logged in the beta database; partnered with the founder on accelerator applications.
Next.jsTypeScriptAWSSupabase
Epic— Care Everywhere — healthcare interoperability platform
Sep 2019 – Jul 2022Software Engineering Manager (R&D Team Lead)
Madison, WI- Promoted from IC to manager (3 direct reports), then to Medium Team Lead (5 direct reports) within a year; one of two team leads in Care Everywhere's Results division (labs and imaging, ~50 people), setting roadmap and coordinating across adjacent teams.
- Led the effort to rebuild Aura for scale after its initial launch — one of the largest projects in Care Everywhere and still referenced in Epic staff meetings; contributed directly to sales proposals for early customers who became paying customers, and represented the platform in sales calls and on-site hospital visits.
- As IC, integrated with legacy hospital systems and complex healthcare interoperability standards (HL7/CDA-era) to scale patient-record exchange to 250M+ records/month; modernized a legacy VB codebase to React and built a clinical training feature that became Epic's top-performing educational tool of the year by views.
ReactC#
University of Virginia— Teachman Lab / Barnes Lab, Dept. of Psychology
Aug 2018 – Jun 2024Machine Learning Researcher
Charlottesville, VA- Built ML models (random forest, logistic regression) for predicting early dropout from web-based cognitive bias interventions for anxiety; F1 ~0.92 across ~1,200 participants.
- Co-authored two peer-reviewed publications: Baee et al., JMIR Mental Health (2024); Eberle et al., Journal of Consulting and Clinical Psychology (APA, 2024). Presented at ABCT 2024.
Pythonscikit-learnRandom forestLogistic regression
Education
University of Virginia
B.A. in Computer Science — awarded High Distinctions
Skills
- AI / LLM
- Claude Code, Codex, MCP, Playwright MCP, RAG, Agentic workflows, Prompt engineering
- Languages
- TypeScript, Python, JavaScript, SQL, C#
- Systems & Tools
- AWS, EC2, Postgres, Supabase, pgVector, React, Next.js, Tailwind, Docker, GitHub Actions
- Practices
- Git version control, code review, CI/CD, testing, observability, legacy system integration
Publications
- Baee et al. (2024). Early Attrition Prediction for Web-Based Interpretation Bias Modification to Reduce Anxious Thinking: A Machine Learning Study. JMIR Mental Health. https://mental.jmir.org/2024/1/e51567
- Eberle, J. W., Daniel, K. E., Baee, S., et al. (2024). Web-Based Interpretation Bias Training to Reduce Anxiety: A Sequential, Multiple-Assignment Randomized Trial. Journal of Consulting and Clinical Psychology. https://doi.org/10.1037/ccp0000896