AI Engineer

Hi, I'm Ernesto. I build AI systems end to end: RAG, agents and the backend that makes them reliable.

See the work

Incoming query

Who is Ernesto and what has he built?

Five repositories retrieved

Query expanded to

Read the case studies

  1. Featured

    Ring TelemetryOps

    Diagnostic agent for Ring video doorbells, built for the Amazon Developer Hackathon 2026. OpenCV measures the stream, a deterministic pre-check sets the severity, and Claude Haiku 4.5 on Amazon Bedrock reasons over the evidence. A local fallback keeps the same report contract when Bedrock is unavailable.

    score 1.000

    matched: agents, RAG, LLM, retrieval, computer vision

  2. Hybrid RAG API

    A hybrid RAG engine written from scratch, without LangChain or LlamaIndex. Dense search (pgvector, HNSW) and lexical search (PostgreSQL GIN) are fused with Reciprocal Rank Fusion, and the prompt requires a document-and-page citation for every claim.

    score 0.448

    matched: RAG, LLM, retrieval

  3. WakeMap

    Flutter prototype from my dissertation: location-based alarms with geofencing, voice-created alarms and an AI travel guide. A Node.js/Express proxy keeps the Gemini key on the server.

    score 0.484

    matched: LLM, backend, mobile

  4. Also built

    AI Surveillance Hub

    An Arduino proximity trigger wakes a Raspberry Pi camera that runs YOLOv8, sends a Telegram photo alert and logs each detection to a Flask dashboard.

    score 0.429

    matched: computer vision, backend

  5. Secure Auth Go

    Go web app for gym membership records, built for a web security course: bcrypt, server-side sessions, CSRF tokens, CSP headers, role-based access and login rate limiting.

    score 0.407

    matched: backend, security

Three systems, from the inside

What each one solves and the decisions that shaped it. The details live in each repository.

My stack

AI and retrieval

  • Hybrid RAG (HNSW, GIN, RRF)
  • Agentic workflows
  • Claude on Amazon Bedrock
  • OpenAI API
  • Google Gemini
  • Structured prompting
  • OpenCV
  • YOLOv8

Backend and data

  • Python
  • FastAPI
  • Flask
  • Node.js
  • Express
  • Go
  • Java
  • Spring Boot
  • PostgreSQL + pgvector
  • Snowflake
  • SQLite
  • MySQL

Cloud and deployment

  • Docker
  • AWS
  • Microsoft Azure
  • CI/CD
  • Git
  • Linux

Frontend and mobile

  • TypeScript
  • React
  • Next.js
  • Tailwind CSS
  • Flutter
  • Dart

The answer, in short

I'm Ernesto, an engineer who builds AI systems end to end: the retrieval, the agent, the backend around them and the tests that say what still breaks.

I graduated from York St John University in the UK with a First Class BSc (Hons), after a Higher Technician diploma in multiplatform app development in Spain. During my internship at CEU's IT Innovation department I worked on an AI tutor that connected institutional data in Snowflake with OpenAI models on Azure.

I'm based in Madrid, available now and open to relocation or hybrid work. I'm looking for AI Engineer, Forward Deployed Engineer and related roles.

Experience

AI Software Developer Intern

CEU, IT Innovation Department

Feb 2025 – Jun 2025

  • Built an AI personal tutor that links institutional data in Snowflake with OpenAI APIs on Azure.
  • Optimised SQL joins, taking data retrieval from days to near-instant.
  • Used structured XML prompting to make outputs more deterministic.

Education

  • BSc (Hons), First Class Honours

    York St John University, United Kingdom

  • Higher Technician in Multiplatform App Development (DAM)

    Spain

Languages

  • Spanish, native
  • English, degree completed in the UK