# Jaime Aza — AI Developer & Full-stack Engineer

> Full-stack software engineer focused on AI-powered products. Experience with LLMs, RAG systems, embeddings, vector databases (Pinecone, Qdrant, pgvector/Supabase) and intelligent agents built with LangChain, LangGraph and MCP. Orchestrates AI agents (Claude Code, subagents, skills, CLIs) to accelerate the development of real products.

- **Name:** Jaime Aza
- **Role:** AI Developer & Full-stack Engineer
- **Location:** Colombia (remote work)
- **Website:** https://jaimeaza.tech
- **Spanish version:** https://jaimeaza.tech/llms.txt
- **LinkedIn:** https://www.linkedin.com/in/jaimeaza/
- **GitHub:** https://github.com/Jjat00
- **Contact (WhatsApp):** https://wa.me/573164277879

## About me

My AI-first approach combines backend and frontend development, automation and real product design. I work with technologies such as React, Node.js, TypeScript, Python, FastAPI, Django REST Framework and n8n, building robust architectures, RESTful APIs, multi-tenant systems and automated flows connected to platforms like WhatsApp or Telegram. I use generative AI tools such as Cursor for AI-assisted development, Lovable and v0 for rapid interface prototyping, and ElevenLabs for voice synthesis in conversational applications.

I have built intelligent agents integrating complex conversational flows, business logic and contextual memory. I am self-taught, curious and product-oriented: I love learning, experimenting with new technologies and building functional solutions with real impact, leading projects from idea to deployment.

## Capabilities

- **AI Agents** — Intelligent agents with LangGraph, LangChain, AI SDK, Skills, CLI and MCP to automate complex processes.
- **Chatbots & Conversational AI** — Chatbots integrated with the WhatsApp API, Telegram and voice with ElevenLabs.
- **Full-stack Development** — Web apps with React, Next.js, Python, TypeScript, Node.js, Django and robust REST APIs.
- **RAG & Vector Search** — RAG systems with embeddings, pgvector, Qdrant and semantic search.
- **Computer Vision** — Image processing with OpenCV, CLIP, DeepFace and PyTorch.
- **AI Agent Orchestration** — Accelerated development by orchestrating AI agents with Claude Code, subagents, skills, MCP and CLIs.

## Main stack

LangChain, LangGraph, AI SDK, Skills, CLI, MCP, Python, React, TypeScript, Supabase, RAG, Claude Code, WhatsApp API, Next.js.

## Achievements and recognitions

- **Winner of Platanus Hack 26 — Buenos Aires (May 2026).** First place in the *AI Security* track and *overall winner* of the Platanus hackathon in Buenos Aires — 36 hours non-stop, 110 hackers selected from across LATAM, with Anthropic as main sponsor. I built *Tranquera*, a proxy interceptor for Claude Code that enforces organizational policies at runtime. Event: https://hack.platan.us/26-ar
- **Winner of the Colombia Tech Week 2025 AI Hackathon (August 2025).** First place in the *Best Agentic App in WhatsApp* track of Colombia's largest hackathon, organized by Colombia Tech Week in Bogotá with over 150 participants. Event: https://www.colombiatechfest.ai-hackathon.co/

## Stats

- 5+ years of experience
- 2x hackathons won
- 3 startups supported

## Professional experience

- **AI Developer Engineer — Ungga** (Jun 2025 – Present, Remote, Proptech Mexico). AI agents with LangGraph to automate complex real estate processes. Chatbots integrated with the WhatsApp API for real-time customer support. Stack: Python, TypeScript, LangGraph, LangChain, WhatsApp API.
- **Full-stack Developer — Higher Bit Solutions** (Mar 2025 – Jun 2025, Remote, Chilean Startup). Web apps with React + TypeScript and REST APIs with Django REST Framework. Stack: React.js, Python, Django REST, TypeScript, Tailwind CSS.
- **Full Stack Developer Engineer — Decimetrix** (Jan 2022 – Feb 2025, Colombia). Led the development team; refactored the frontend, cutting the bundle size by 50%. React + Mapbox modules that improved field data collection by 40%. Stack: React.js, Node.js, PostgreSQL, Mapbox, Docker, AWS.
- **Software Development Engineer — Decimetrix** (May 2021 – Dec 2021, Colombia). MVP of Green Dragon, a digital twin for carbon footprint management in oil & gas. Stack: React.js, Node.js, PostgreSQL, AWS, Docker.
- **Computer Vision Engineer — Intecol SAS** (Jan 2021 – Apr 2021, Colombia). Desktop app for FLIR/Cognex industrial cameras; sped up acquisition by 30%. Stack: Python, OpenCV, PyQt.

## Projects

### Tranquera
Proxy interceptor for Claude Code that closes the organizational alignment gap: Claude Code is aligned with Anthropic's values, but not with the policies of each company deploying it. Tranquera is the middle layer — a silent customs checkpoint that enforces no-code rules at runtime with a Regex → Pattern → Haiku 4.5 judge cascade under 200ms of overhead and four explicit actions (BLOCK · REDACT · WARN · LOG). The (non-technical) compliance officer defines policies in natural language from a visual builder; devs onboard with a single command (`npx tranquera setup`). Includes an AI Suggestor with a daily cron and Haiku 4.5 that proposes new rules from the LOGs with a human in the loop. Postgres + pgvector to pre-filter rules by semantic similarity. Winner of the AI Security track and the overall prize at Platanus Hack 26 Buenos Aires (main sponsor: Anthropic).
- Technologies: Claude Haiku 4.5, AI Security, Prompt caching, Regex → Pattern → LLM cascade, Next.js 16, FastAPI, Python 3.12, Prisma, PostgreSQL, pgvector, Auth.js, Google OAuth, Vercel Cron, TypeScript, Tailwind CSS, Turborepo
- Web: https://tranquera.vercel.app/
- Repository: https://github.com/platanus-hack/platanus-hack-26-ar-team-22

### Tresqu
A team of financial agents you talk to via chat — from WhatsApp, Telegram, Gmail or the web. Its promise: "it knows how you live, it invests like you are." Four specialized agents coordinate: (1) Expenses & income records and classifies transactions; (2) Investments prepares stock operations on Wallbit but never executes them without your approval; (3) Market analyst contextualizes prices and fundamentals according to your profile; (4) Risk profile measures your real tolerance from your history and blocks incongruent operations. You log expenses the way you talk — text, audio or a photo of the receipt — with Gmail sync and natural language search over embeddings in pgvector. The money lives in Wallbit (not in Tresqu): buy and sell stocks, ETFs and bonds with multi-currency support (COP, USD, EUR) and real-time conversions.
- Technologies: Multi-agent system, LangChain, OpenAI, RAG, pgvector, NLP, Voice and image (receipts), WhatsApp Business API, Telegram Bot API, Gmail API, Wallbit API, Web Speech API, Django 5.2, DRF, PostgreSQL, React, TypeScript, shadcn/ui, Tailwind CSS, Zustand, JWT
- Web: https://tresqu.com/
- Frontend: https://github.com/Jjat00/tresqu-frontend
- Backend: https://github.com/Jjat00/tresqu-api

### Festora
Professional gallery delivery platform for photographers with multimodal AI. It analyzes every photo with GPT-4o and Gemini 2.0 Flash: assigns a 1-10 score, evaluates composition, pose and background quality, categorizes scenes and generates highlights. Includes semantic photo search in natural language with multimodal RAG using Gemini Embedding 2 (text and image in the same vector space) stored in pgvector. The public gallery reorders photos based on the client's favorites and writes poetic cover phrases with AI. Complemented by festora-vision-api: a microservice with blur detection (Laplacian), BRISQUE/NIMA technical scoring, emotion analysis (DeepFace) and CLIP/DBSCAN semantic clustering.
- Technologies: GPT-4o, Gemini 2.0 Flash, Gemini Embedding 2, Multimodal RAG, Vercel AI SDK, CLIP, DeepFace, Computer Vision, PyTorch, LiteLLM, pgvector, Next.js, Prisma, PostgreSQL, FastAPI, Cloudflare R2, Auth.js, Docker
- Web: https://festora-gamma.vercel.app/
- App: https://github.com/Jjat00/festora
- Vision API: https://github.com/Jjat00/festora-vision-api

### Frostbyte
End-to-end platform for beverage businesses (slushies, frappés, cocktails, bars and restaurants) with native AI at every touchpoint. The digital menu recommends drinks based on the customer's mood via GPT-4o-mini, transcribes voice commands with Whisper and shows daily motivational phrases. The admin panel includes a professional product image generator with GPT Image 1.5 and Cloudflare R2, automatic description suggestions, per-table order management, inventory, operating expenses, analytics and a real-time multiplayer game 'Duelo Frostbyte' over WebSocket.
- Technologies: GPT-4o-mini, GPT Image 1.5, Whisper, Generative AI, WebSocket, Real-time, React 18, Vite, Tailwind CSS v4, Zustand, TanStack Query, Framer Motion, Django 6, DRF, PostgreSQL, Redis, Django Channels, Cloudflare R2, Recharts
- Web: https://frostbyte.com.co/landing
- Frontend: https://github.com/Jjat00/frostbyte-frontend
- Backend: https://github.com/Jjat00/frostbyte-backend

### Keyduelo
Real-time multiplayer typing race inspired by Monkeytype. Anyone can create a room with a 5-character code, share the link and compete for the highest WPM on the same text, with spectator mode. The architecture runs 100% on the edge over Cloudflare Durable Objects: one 'Room' DO per room (authoritative state, WebSockets with the Hibernation API, persistent hostToken) and a 'RoomRegistry' singleton that keeps the global list of active rooms via RPC between DOs. The clock, the text and the ranking are decided server-side (cheating is impossible). Includes 5 themes, sounds synthesized with the Web Audio API and a host/player/spectator model.
- Technologies: Cloudflare Workers, Durable Objects, WebSocket Hibernation API, Edge Computing, Real-time, Next.js 16, React 19, TypeScript 6, Tailwind CSS v4, Turborepo, pnpm, Web Audio API, Vercel
- Web: https://keyduelo.vercel.app/
- Repository: https://github.com/Jjat00/keyduelo

### S2T — Speech to Text for Windows
Windows desktop app that transcribes speech to text in real time and types the result directly wherever the cursor is — no copying, no pasting. Push-to-talk with F9. Supports two swappable engines: Deepgram Nova-3 (cloud, low latency, simultaneous es+en bilingual) or local faster-whisper (no internet, no API key). Includes a floating HUD with timer, live preview, voice visualizer and similarity-based deduplication. Lives as a system tray icon, with VAD via silero-vad ONNX. Packaged with PyInstaller + Inno Setup as a native .exe installer.
- Technologies: Deepgram Nova-3, faster-whisper, silero-vad ONNX, Speech-to-Text, Push-to-talk, Real-time, Python 3.12, uv, PyAudio, pynput, pystray, tkinter, PyInstaller, Inno Setup, Windows
- Repository: https://github.com/Jjat00/s2t-windows

### image_to_xyz
Web app that converts a 2D image into a navigable 3D point cloud using AI depth estimation. The backend (FastAPI + Docker on Railway) wraps Depth-Anything-V2 (ViT-S checkpoint) and returns the depth map as a grayscale PNG. The frontend combines the RGB image and the depth map to generate one 3D point per pixel and render it with React Three Fiber + drei as a THREE.Points with vertexColors. Supports three swappable depth backends: a self-hosted FastAPI server, HuggingFace Spaces or the OpenAI Images API.
- Technologies: Depth-Anything-V2, Computer Vision, Generative AI, Three.js, React Three Fiber, WebGL, Point Cloud, FastAPI, PyTorch, Docker, OpenCV, React, Vite, TypeScript, Railway, Vercel
- Web: https://image-to-xyz.vercel.app/
- Repository: https://github.com/Jjat00/image_to_xyz

### RAG Properties
Real estate semantic search engine with a RAG architecture covering more than 8,800 properties in Mexico. The user asks in natural language, Gemini 3 Flash extracts structured filters (city, type, bedrooms, price), Qdrant pre-filters and vector search retrieves the most relevant properties. Supports multimodal RAG with Gemini Embedding 2: each property is represented by a single 3072-dimension vector fusing the text description with up to 6 images in the same vector space. Supports swappable embeddings (OpenAI text-embedding-3 and gemini-embedding-001) with a React playground featuring an interactive similarity chart and a conversational agent with LangGraph (ReAct).
- Technologies: Multimodal RAG, Gemini Embedding 2 (3072d), OpenAI text-embedding-3, Gemini 3 Flash, LangGraph ReAct, Qdrant, Vector Search, NLP, FastAPI, Python 3.12, React 19, Vite, shadcn/ui, Tailwind v4, Railway, Vercel
- Web: https://rag-properties.vercel.app/
- Repository: https://github.com/Jjat00/rag-properties

## Frequently asked questions

**Who is Jaime Aza?**
Jaime Aza is a Colombian full-stack software engineer (AI Developer & Full-stack Engineer) specialized in AI-powered products. He builds AI agents, RAG systems and web applications with LangGraph, LangChain, MCP, React, Next.js and Python. He is a winner of Platanus Hack 26 (Buenos Aires) and the Colombia Tech Week 2025 AI Hackathon.

**What does Jaime Aza specialize in?**
Applied AI: intelligent agents with LangGraph, LangChain, AI SDK and MCP; RAG systems with embeddings and vector search (pgvector, Qdrant); conversational chatbots (WhatsApp, Telegram, voice with ElevenLabs); computer vision; and full-stack development with React, Next.js, TypeScript, Python, FastAPI and Django.

**Which hackathons has Jaime Aza won?**
He won Platanus Hack 26 in Buenos Aires (May 2026): first place in the AI Security track and overall winner, with Anthropic as main sponsor. He also won the Colombia Tech Week 2025 AI Hackathon (August 2025) in the Best Agentic App in WhatsApp track.

**Where does Jaime Aza work?**
He works remotely as an AI Developer Engineer at Ungga, a proptech company in Mexico, building AI agents with LangGraph and chatbots with the WhatsApp API. He previously worked at Higher Bit Solutions, Decimetrix and Intecol SAS.

**How can I contact Jaime Aza?**
Via LinkedIn (linkedin.com/in/jaimeaza), GitHub (github.com/Jjat00) or WhatsApp (wa.me/573164277879). His official website is https://jaimeaza.tech.

## Contact

- LinkedIn: https://www.linkedin.com/in/jaimeaza/
- GitHub: https://github.com/Jjat00
- WhatsApp: https://wa.me/573164277879