In the product
LLM features people actually touch: conversational shopping, AI image generation, custom GPTs. Streamed over SSE and WebSockets so the interface responds in real time instead of hiding behind a spinner.
Senior full-stack engineer · AI products
I build AI products people actually use.
Lately: conversational shopping assistants.
I'm Jeremy, a senior full-stack engineer from Santo Domingo with 8+ years shipping web products. Lately that means LLM features in production — on a React / Next.js and TypeScript front end, with Node.js (NestJS) and Python behind it.
AI where it counts — and the full stack to ship it.
In the data
Models that read and sort: TensorFlow computer vision on handwritten medical documents, Pandas classification, and analytics pipelines that turn AI conversations into topics merchants can act on.
In how I build
Claude Code and Cursor every day, with MCP connected to Linear and GitHub so the assistant works from real tickets and repo context — faster delivery without lowering the bar.
AI products I helped ship.
01AISwap Commerce · Paul Smith
Live AI virtual try-on
Avatar creation with photo capture and age verification, AI image-generation jobs streamed live to the browser, and secure private-media delivery on a React 19 server-rendered app.
02AISwap Commerce
Conversational shopping assistant
The shopping experience around the AI assistant — wardrobe and bag flows, buy-the-outfit journeys, multi-currency — updated in real time over WebSockets and SSE.
03AITruelogic · MileIQ
Custom GPTs & chatbots
A custom GPT on OpenAI and Anthropic models that processes and analyzes company-specific data, plus an OpenAI-powered chatbot and Pandas data classification at MileIQ.
04AIGrupo Humano
AI medical-document kiosk
A React kiosk that read handwritten medical documents with TensorFlow computer-vision models, backed by Node.js and Spring Boot APIs.
05Swap Commerce · Truelogic
Full-stack foundations
What the AI runs on: a merchant analytics pipeline (React → WebSockets → FastAPI ETL → PostgreSQL → dashboards), a multi-currency FX subsystem shipped to GKE with Helm and ArgoCD, OAuth hardening against CSRF and session fixation, and Storybook design systems.
Stack
Pick a layer.
// Shipping LLM features to real users
- LLM features
- OpenAI
- Anthropic
- Claude Code
- Cursor
- MCP
- TensorFlow
// Where I spend most of my time
- React
- Next.js
- React Native
- Redux
- Zustand
- Tailwind
- Material UI
- Storybook
- D3
- WCAG
// APIs and real-time transport
- Node.js
- NestJS
- Express
- FastAPI
- Apollo GraphQL
- REST
- WebSockets
- SSE
// Storage, containers and clusters
- PostgreSQL
- MongoDB
- Redis
- Docker
- Kubernetes
- Helm
- AWS
- GCP
// Daily drivers
- TypeScript
- JavaScript
- Python
// Tests and pipelines
- Cypress
- React Testing Library
- Jest
- Vitest
- GitHub Actions
- Jenkins
Experience
The road so far.
Filter by technology, then open a stop on the timeline.
-
- Built the live AI virtual try-on: avatar creation with photo capture and age verification, AI image-generation jobs streamed live to the browser, and secure private-media delivery on a React 19 SSR app.
- Developed the shopping experience around the AI assistant — wardrobe and bag flows, buy-the-outfit journeys, multi-locale / multi-currency — driven in real time by WebSockets and SSE.
- Owned storefront auth end to end (email OTP + Google OAuth with cart-restoring sessions) and closed an OAuth CSRF / session-fixation hole with nonce-based state verification.
- Owned merchant analytics across the stack: React instrumentation → WebSocket ingestion → Python / FastAPI ETL → PostgreSQL → REST metrics APIs → dashboards.
- Shipped a multi-currency FX subsystem to GKE via Helm and ArgoCD; root-caused a launch-week dropped-events incident and hardened delivery with a buffered outbox.
- Built the UI foundation in Storybook and rebuilt admin tooling, replacing 4,500+ lines of legacy code.
- React 19
- TypeScript
- Node.js/NestJS
- Python/FastAPI
- WebSockets
- SSE
- PostgreSQL
- Redis
- GCP
- GKE
- Helm
- ArgoCD
- LLMs
-
- Built a high-traffic streaming platform on Next.js with a versatile video player for live streams, movies and on-demand shows.
- Delivered a React / TypeScript app using Redux and Zustand for state and Tailwind for a consistent, responsive UI.
- Built in-app analytics to track and visualize user interactions; partnered with design in Figma on a reusable component library.
- Next.js
- React
- TypeScript
- Redux
- Zustand
- Tailwind
- Figma
-
- Designed an optimized component architecture for a large-scale, high-concurrency web app.
- Improved SEO for discoverability and search rankings.
- Removed obsolete React code, cutting bundle size by 15%.
- React
- TypeScript
- Next.js
- Tailwind
- Redux
- REST
-
- Designed Contentstack content models — content types, global fields and taxonomies — for a modular content architecture.
- Built React interfaces backed by the headless CMS for flexible, content-driven pages.
- React
- Contentstack
- TypeScript
-
- Designed and deployed React / Next.js (12 → 13) apps for enterprise financial workflows, using ISR and the App Router.
- Built a custom GPT solution on OpenAI and Anthropic models to process and analyze company-specific data.
- Built APIs with FastAPI, Apollo GraphQL and REST (Node.js / Express) on PostgreSQL; ran workloads on Kubernetes with Helm.
- Designed a React design system documented in Storybook, integrating Material UI and Tailwind.
- React
- TypeScript
- Next.js
- FastAPI
- Apollo GraphQL
- Node.js
- Python
- PostgreSQL
- Docker
- Kubernetes
- Helm
- OpenAI
- Anthropic
-
- Built a mobile-responsive mileage-tracking app on Next.js 14 with React Server Components.
- Led AI initiatives in Python: an OpenAI-powered chatbot and data classification with Pandas.
- Migrated the mapping stack from Bing Maps to Google Maps, including routing and directions.
- React
- Next.js
- TypeScript
- Node.js/NestJS
- Python
- GraphQL
- OpenAI
- Pandas
- Cypress
-
- Migrated a large codebase from Webpack 4 to 5, introducing Module Federation to share components across apps.
- Moved a large React codebase from classes to hooks and added useTransition / Suspense UX improvements.
- Implemented AWS Cognito authentication.
- React
- TypeScript
- Material UI
- Webpack
- Node.js
- RTL
- AWS
-
- Architected a React app powering an AI-driven kiosk that processed handwritten medical documents with TensorFlow computer-vision models.
- Delivered REST APIs across Node.js / Express and Java 8 / Spring Boot.
- Built Jenkins CI/CD pipelines; used React Native with Redux for mobile features.
- React
- React Native
- Redux
- TensorFlow
- Node.js
- Express
- Java 8
- Spring Boot
- Jenkins
- AWS
-
2012 – 2018 · Education B.Sc. Software Engineering — UNAPEC Associate, Software Engineering — ITLA2019 · Certification Scrum Foundation Professional Spanish (native) · English (professional)