felixvalor.com▮
ES EN

[00 / CAPTURE]

detection: full_name · 1.00

Félix
Valor

I build AI systems that turn messy data into reliable information

[ let's talk ] download_cv

open to new opportunities

// class: knight
x:---- y:---- ≈2,500 points · hover me

[01 / INGEST]

About me

Software engineer with 4+ years building and deploying AI solutions — from generative AI (LLMs, RAG, agents, fine-tuning, LangChain) to classic machine learning and computer vision (YOLO, BERT, Whisper). Always taken from idea to production.

I have worked directly with high-profile clients, in contexts where mistakes have consequences: electoral processes for the governments of Argentina, Spain and Catalonia —over 100,000 records processed within hours on voting day—, Real Madrid, Aena, Instituto Cervantes and Plenitude. I also ship my own products solo, from scratch to production.

I move comfortably across backend architecture, MLOps and cloud deployment (Azure, GCP). I am looking to take the next step at a strong company in the field, with a special interest in computer vision projects.

[02 / PROCESSING]

Experience

may 2025 — present

madrid, spain

relevance: 0.98

Software Engineer · Generative AI — BC Digital live

  • 10+ document data extraction projects using generative AI (LangChain and Vertex AI) for the real estate sector, processing over 5,000 documents a day at 90–95% extraction accuracy: property title copies, land registry excerpts, appraisals, invoices, ID documents, cadastral certifications and energy performance certificates, among others.
  • Continuous research into new LLM models and extraction techniques, rolling improvements into live production systems without breaking them.
  • Direct work on production across a multi-repository environment, prioritizing stability and continuous improvement.

jan 2023 — apr 2025

madrid, spain

relevance: 0.95

Software Engineer · AI Focus — Indra

  • Multi-provider generative AI (GPT-4, Gemini, Claude, Mistral) for information extraction and understanding, applying RAG, agentic architectures, advanced prompting and fine-tuning / domain pretraining.
  • Hybrid solutions combining generative AI with classic machine learning and computer vision: object detection in images and video (PyTorch, TensorFlow, YOLO), text processing (BERT), audio transcription (Whisper) and image processing (OCR/ICR).
  • Design and deployment of Python (FastAPI) microservices containerized with Docker on Azure, GCP and OpenShift.
  • National and international projects with high-profile clients: electoral processes (governments of Argentina, Spain and Catalonia), Real Madrid, Aena, Instituto Cervantes and Plenitude, with availability for extended on-site travel.

sep 2022 — dec 2022

dublin, ireland

relevance: 0.92

Software Engineer (Erasmus+) — 4Property

  • Python web scraping and maintenance of SQL databases on Azure.
  • Hands-on introduction to ElasticSearch, Grafana and Apache Kafka.

mar 2022 — jun 2022

ciudad real, spain

relevance: 0.89

Software Engineer — Indra

  • Computer vision and image processing projects: ID card number recognition from photographs and digit extraction from electoral process records.

[03 / DETECTIONS]

Featured projects

A selection of the work that best represents what I do.

detection: electoral_processes

Electoral Processes

2023 — 2024

Semi-automatic system to analyze scanned electoral records and assign each party’s votes, adding a verification layer on top of the manual (BPO) work. Deployed in Argentina 2023, Catalonia 2024 and the 2024 European Elections. I developed the whole codebase and helped design the solution. On election day we set up and configured the machines on site and monitored every Kubernetes instance live throughout the count.

In Argentina we processed over 100,000 records within hours on voting day, at 90%+ accuracy in the latest iteration. The manual review team cleared the bulk of the count in a fraction of the time previous elections had required, and the verification layer kept human errors out of the final tally.

pythonyolotensorflowopencvkubernetesgoogle clouddocker
Confidential
detection: chessscan_2_0

ChessScan 2.0

2025 — 2026

Full rebuild of my original capstone project: two chained YOLOv8 models —board segmentation, from whose mask I derive the 4 corners and the homography, plus 12-class piece detection— trained on a custom dataset of ~130 images and 2,400 hand-labeled boxes (Label Studio + Ultralytics). A sweep of 28 training runs (YOLOv8 vs YOLO11, nano to x) confirmed the nano model was enough, keeping inference cheap on Cloud Run. React / TypeScript / Vite / Tailwind frontend.

pythonyoloultralyticsfastapireacttypescriptcloud rundocker
see_it_live how_it_started
detection: smart_email_classification_real_madrid

Smart email classification · Real Madrid

2023 — 2025

Automatic classification of the Real Madrid members’ support inbox using natural language processing, at 95%+ accuracy. It brought order to an inbox that until then had no classification of any kind, and was integrated alongside the Power Automate team so the output could be read at a glance.

pythonnlpllmspower automate
Confidential
detection: docsutils

docsutils

2025 — 2026

PDF tools platform built with FastAPI and Docker, deployed on GCP Cloud Run, with self-hosted analytics (Umami). Built solo from scratch to production: product, backend, deployment and domain.

pythonfastapidockercloud runumami
see_it_live
detection: claude_code_multi_repo_setup

Claude Code multi-repo setup

2025 — 2026

Shared conventions and reusable commands to streamline development and incident triage across multi-repository environments.

claude codeautomationdx

[04 / CLASS_MAP]

Tech stack

Tools and technologies I work with every day.

[04.1] generative_ai_&_llms

  • 01.01 GPT-4, Gemini, Claude, Mistral
  • 01.02 LangChain
  • 01.03 Vertex AI
  • 01.04 RAG
  • 01.05 Agentic architectures
  • 01.06 Fine-tuning / domain pretraining
  • 01.07 Advanced prompting

[04.2] machine_learning_&_computer_vision

  • 02.01 PyTorch
  • 02.02 TensorFlow
  • 02.03 Scikit-Learn
  • 02.04 pandas
  • 02.05 YOLO
  • 02.06 BERT
  • 02.07 Whisper
  • 02.08 OpenCV
  • 02.09 MLOps
  • 02.10 OCR / ICR

[04.3] backend

  • 03.01 Python (FastAPI, Flask)
  • 03.02 Java
  • 03.03 .NET
  • 03.04 Microservices architecture

[04.4] cloud_&_devops

  • 04.01 Docker
  • 04.02 Kubernetes
  • 04.03 GCP Cloud Run
  • 04.04 Azure
  • 04.05 OpenShift
  • 04.06 CI/CD

[04.5] data_&_frontend

  • 05.01 SQL
  • 05.02 ElasticSearch
  • 05.03 Apache Kafka
  • 05.04 Grafana
  • 05.05 React
  • 05.06 TypeScript
  • 05.07 Vite

[05 / METADATA]

Education

degree
Cross-platform Application Development (DAM)
school
IES Gregorio Prieto · 2020 — 2022 · Valdepeñas, Spain
spanish
native
english
c1 comprehension · b2 production

[06 / OUTPUT]

Let's talk

Open to new opportunities in the AI field. Drop me a line and I’ll get back to you soon.

POST /email CALL /phone GET /linkedin GET /github

// all endpoints respond — avg time: <24h · madrid, spain · remote