Full-Stack Engineer —
React/TypeScript · Node/Postgres

Bootstrapped and shipped a production AI content platform end-to-end with auth, usage-based billing, and streaming UX.

Problem
Writers and marketers needed a faster way to generate on-brand content with transparent usage tracking and predictable billing.

Actions

• Designed and built the product end-to-end: React/TypeScript app, Node/Express APIs, PostgreSQL schema.

• Implemented Auth0 (OAuth2/JWT) with RBAC and secure session flows.

• Built usage-based billing (Stripe) with credits, metering, idempotent webhooks, and proration handling.

• Added Server-Sent Events for token-streaming responses; optimized perceived latency.

• Deployed on Cloud Run + Cloudflare Pages; added observability (structured logs, basic tracing), CI/CD.

• Wrote OpenAPI specs and request validation to reduce integration errors.

Built most of the frontend and a significant slice of the backend for a data preprocessing/orchestration app used by internal teams.

Problem
Bio/ops teams needed a reliable way to ingest datasets, orchestrate preprocessing, and review results in a single UI.

Actions

• Built ~80% frontend (React/TypeScript) and ~50% backend (Node APIs) with consistent patterns and shared components.

• Implemented Auth0 tenant with role-based access; integrated Chargebee for account/billing workflows.

• Standardized API contracts via OpenAPI; added request validation and error envelopes.

• Set up builds/deploys on AWS/DigitalOcean; added lightweight CI and release notes.

• Partnered with design + bio teams; shipped weekly iterations.

Rebuilt a slow/unstable analytics UI into a fast, memory-efficient app that handled large data interactively.

Problem
The original Next.js + Plotly app struggled with large datasets (multi-second interactions, memory spikes, crashes).

Actions

• Rebuilt ~90% of the frontend (Next.js + Plotly), introducing data virtualization, memoization, and chunked parsing.

• Implemented compute/memory optimizations (e.g., downsampling, worker offloading where helpful).

• Introduced error boundaries and resilient state management; added targeted unit/UI tests.

• Instrumented key interactions to track latency and memory footprint over time.

• Personally designed the UI from scratch.

Results

• Reduced interaction latency from multi-seconds → near-instant for typical views.

• Supported tens of MB per session reliably.

Built the entire frontend and half of the backend for a lab tool that coordinated multi-device study sessions in real time.

Problem
Lab teams needed synchronized timers, device telemetry, and error-resilient workflows across multiple local machines.

Actions

• Built 100% of the frontend (React/TypeScript) and ~50% Node/Python backend endpoints.

• Integrated MQTT telemetry; implemented robust timer orchestration and cross-device sync.

• Designed error-tolerant UX (retries, offline/restore flows, session resume).

• Added logging and session audits to investigate anomalies and improve reliability.

• Designed the full UI from scratch according to customer needs.