Personal details
Jarvis A.
Technical skills
About
Staff/Principal Full-Stack & AI Platform Engineer — TypeScript, Node.js, React/Next.js, PostgreSQL, AWS, production LLM systems. 25+ years shipping cloud-native SaaS, payment systems, and AI-integrated products. Most recently designed and built BudgetBites, a four-service strict-TypeScript platform (Next.js 15, Express API gateway, Playwright ingestion, Prisma/Auth.js/Stripe) now live on AWS and taking paid subscriptions.
I work in the codebase: architecture, implementation, PR review, query optimization, auth and payment hardening, deployment. Uncommon pairing of engineering depth and interaction-design fluency — I take ambiguous problems through spec, user flows, architecture, and shipped code. Disciplined use of AI coding agents, with human ownership of system design, correctness, and security.
Available for full-time Staff/Principal roles and senior architecture/build engagements. Existing ventures are maintenance-mode and do not compete for attention.
Work Experience
- Designed and built a four-service strict-TypeScript platform: Next.js 15 / React 19 consumer app (Auth.js, Prisma, Stripe subscriptions), Express API gateway as the sole trust boundary for pricing data and AI calls, Dockerized Playwright/Cheerio ingestion of Publix, Kroger, and Aldi circulars, and a React/Vite operator console.
- Architecture rule "rules compute, models explain": a deterministic engine owns pricing, macros, and health constraints; Claude classifies and writes narrative only after computation, with a full audit trail.
- Schema-per-app PostgreSQL isolation; x-api-key BFF auth with a reserved per-clinic key path for a B2B clinic pilot.
- Deployed on AWS EC2/PM2 with S3 asset delivery; Sentry and Prometheus on the ingestion stack; Vitest across services.
- Shipped in roughly five weeks via specification-driven, agent-assisted implementation with human architectural review.
- Built a multi-timeframe trading-analysis engine serving ~350 active traders, combining custom liquidity models, Wyckoff logic, and LLM narrative interpretation.
- Node.js + AWS backend with real-time market-data processing and automated deployment; delivers trade plans in under 90 seconds across equities, futures, and crypto to Discord and web clients.
- Built a multi-tenant white-label reseller tier on the same platform.
- Own reliability, observability, and iteration of the live system.