Daniel
Villera
Full-stack developer building AI products end to end: the interface, the server behind it and the data it runs on.
Experience
Where I have worked, most recent first.
- Nov 2025 – Present
Unozero.ai
Co-Founder · Remote
- Built a conversational booking engine on advanced RAG (Gemini, LangGraph, Qdrant) that handles booking CRUD and cut the sales team’s operational load by 75%.
- Led the frontend refactor in Next.js and TypeScript: faster loads and a cleaner UI/UX in the flows that matter most.
- Automated lead classification with AI. The agents qualify potential customers and complete the booking, closing the sale without human intervention.
Next.js · TypeScript · LangGraph · Gemini · Qdrant
- Mar 2023 – Feb 2024
Softpymes
Software Engineer · Cali, Colombia
- Built interactive modules for “Pedidos+”, a B2B2C e-commerce platform, with Angular and Azure, so end customers complete purchases faster.
- Turned business requirements into working features with the backend team, consuming RESTful APIs.
Angular · Azure · REST APIs
Projects
Each one sits at the altitude where it was hardest.
- 2,600 m
BoxBox
A Formula 1 race predictor, end to end. An XGBoost model, validated race by race and retrained after every Grand Prix, gives each driver a win and podium chance, with a SHAP breakdown of why. A FastAPI backend serves it to a bilingual Next.js dashboard.
Visit project ↗
Python · XGBoost · FastAPI · Next.js · PostgreSQL2026 - 2,300 m
Otro Lunes
A Next.js app that turns seven answers into a meal plan and a training routine for every day you train, written by an AI model (Gemini or OpenAI, your choice). The model call runs in a Route Handler and the plan stays in your browser, so there are no accounts and no database.
Visit project ↗
Next.js · TypeScript · React · Zod · Gemini · OpenAI2023
Blog
Notes from the climb, in English and Spanish.
Serverless is not only on demand. It forces you to be stateless
The theory sells serverless as pay-per-use. In practice the bigger change is that your app scales horizontally whether you plan for it or not.
An AI plan with no accounts and no database
How Otro Lunes turns seven answers into a meal plan and a training routine, and why the plan never leaves your browser.
Why BoxBox retrains after every Grand Prix
A win chance without a reason is hard to trust. How BoxBox validates, retrains and explains each prediction.
About
Full-stack developer focused on AI systems (RAG, agents, LLMs) and business intelligence. I work in TypeScript (Next.js, Angular) and Python (FastAPI, LangGraph).
I model data in PostgreSQL and PowerBI and deploy it to the cloud, to automate operations and take friction out of the workflow.

Frontend and UI
TypeScript, JavaScript, Next.js and Angular.
Backend and AI
Python, FastAPI, LangChain, LangGraph, Vercel AI SDK and Qdrant.
Data and cloud
PostgreSQL, Supabase, PowerBI, Azure, AWS, GCP, Git and Docker.
Education
Software Engineering
Universidad Cooperativa de Colombia
2025 – 2027 (expected)
Software Development Technology
Universidad Nacional Abierta y a Distancia
2021 – 2024
Down to sea level
Open to freelance work. Have a project or a question? Write to me.