Juan – Python, LLM, LangChain

A voice-first AI healthcare assistant that helps parents of neurodivergent children across the United States find local resources, screening tools, and support through phone, SMS, and web chat. It guides families through complex resource navigation in plain language, performs location-aware searches for nearby services, and runs on a HIPAA-compliant, low-latency architecture so caregivers can get help even without reliable internet access.

• Built a phone-accessible AI agent with Twilio, AWS, Anthropic Claude, and OpenAI that guides parents through complex resource navigation;
• Integrated location-aware search using AWS RDS/Aurora PostgreSQL with PostGIS, enabling geographic queries for nearby services;
• Implemented Redis caching that cut API costs by 60% while maintaining sub-second response times for common queries;
• Architected multi-turn conversation flows with state machines to handle clarifications and fallback scenarios;
• Designed a HIPAA-compliant, low-latency architecture on AWS (Lambda, API Gateway, WebSockets) following the Well-Architected Framework;
• Built serverless voice-processing endpoints with Fastify, Node.js, and TypeScript;
• Developed the marketing landing page with Next.js, React, and Tailwind CSS, served via S3 and CloudFront;
• Set up CI/CD with GitHub Actions, AWS CodePipeline, and CodeBuild for the agent and landing page.

An AI-powered legal-compliance platform for entrepreneurs and startups that generates state-specific LLC formation documents and streamlines the setup process across English, Spanish, French, German, Italian, and Portuguese markets. It also includes an AI illustrated-book generator that creates custom stories and artwork with an interactive editing experience and print-ready export.

• Built an end-to-end LLC formation platform serving English, Spanish, French, German, Italian, and Portuguese markets, processing 500+ formations in the first 6 months;
• Integrated AI document generation (OpenAI + Puppeteer) for state-specific operating agreements and articles of organization, cutting formation time from 15 minutes to 3;
• Designed a multi-step onboarding flow with real-time validation that reduced formation errors by 85%;
• Built an admin dashboard in React with Redux Toolkit and React Query, plus Stripe billing for formation fees and compliance subscriptions;
• Developed Everbook, an AI illustrated-book generator using GPT-4 for stories and DALL-E for artwork, with an interactive editor and print-ready PDF export;
• Offloaded book generation to SQS and Lambda, cutting average creation time from 15 minutes to under 3;
• Added i18n for 6+ languages with automated content translation via the OpenAI API.

User-friendly Web3 wallets for web and browser extensions, plus a multi-chain NFT portfolio dashboard, built for the ARK blockchain ecosystem. The products focused on making crypto safer and more approachable for non-technical users through clear transaction flows, hardware wallet support, and secure asset management.

• Shipped ARKVault, a web wallet with Ledger hardware support, multi-profile management, and real-time transaction status tracking;
• Refactored the frontend to React 19, cutting bundle size 30% and improving performance by about 50%, backed by 400+ unit tests with Vitest and React Testing Library;
• Built ARKWallet, a Chrome and Firefox extension with BIP44 HD wallet derivation, encrypted key storage, and clear transaction-signing flows;
• Developed Dashbrd, a multi-chain NFT portfolio dashboard integrating Alchemy and Moralis, and migrated styling to Tailwind CSS for a 25% smaller bundle;
• Built ARKConnect, an open-source MetaMask-style dApp-connection extension with secure message signing and approval flows;
• Reduced deployment time by 30% through CI/CD pipeline refactoring and automated testing.

A decentralized sports-betting platform on Solana with a companion Web3 wallet, designed to let users place bets directly on-chain. The work focused on frontend modernization, real-time blockchain integration, and a smoother experience for betting, account management, and secure wallet interactions.

• Refactored 100% of the frontend repositories, cutting size by 40% and deployment time by 20%;
• Modernized the stack to TypeScript, Next.js, Redux, React Query, and Tailwind CSS;
• Built real-time endpoints communicating with the Solana network for live bet placement and settlement, and contributed to the platform SDK and smart contracts;
• Implemented parlays, teasers, and other betting options, and built the full UI for landing, betting, account, profile, and settings;
• Hardened security with geo-IP, 2FA, and multi-layer AWS protections, reducing vulnerabilities by 70%.

An AI-powered talent acquisition product and a high-volume hiring and onboarding platform used by major enterprise clients.

• Designed and built the company’s third product, Advise, from scratch, improving talent-acquisition prediction accuracy by 30%;
• Led a 6-person frontend team across 4 platforms, mentoring engineers and standardizing the stack;
• Built features processing millions of hourly-worker applications for clients like Amazon and Grubhub;
• Migrated frontend repos to TypeScript, Next.js, Tailwind CSS, Redux, and Apollo/GraphQL, and set up CI/CD across 10+ repositories;
• Increased test coverage by 80% with Jest and React Testing Library.

An edtech platform that trained 10,000+ teachers across 6 Latin American countries during the pandemic. It used an adaptive teaching model and an AI chatbot to provide 24/7 support for asynchronous learning and large-scale teacher training.

• Trained 10,000+ teachers across 6 Latin American countries during pandemic school closures;
• Built a scalable 4-tier AWS architecture (EC2, S3, RDS) supporting 5,000+ concurrent users;
• Programmed an AI chatbot that cut response times by 90% and grew revenue by 150%;
• Implemented an adaptive teaching model with AI/ML for 24/7 asynchronous learning;
• Produced 300+ virtual classes and led the full product lifecycle from design to deployment.

A travel coordination platform created during COVID-19 to help international travelers manage flights, vaccine appointments, and lodging for 300+ people. It included an AI monitoring system and a virtual call-center to support complex logistics and keep travel operations organized.

• Designed a 3-tier AWS cloud architecture supporting 3,000+ concurrent users across Peru;
• Built a full-stack platform coordinating flights, vaccine appointments, and lodging for international travelers;
• Implemented an AI monitoring system tracking 300+ patients and platform users;
• Built a virtual telephone-exchange system enabling a 10+ person team to coordinate travel logistics;
• Raised test coverage to 85% across repositories for reliable health-critical operations.