Skip to main content

Hello, I'm

Teo Lorente

I |

Head of AI & Data Engineering — Building intelligent systems that matter

SCROLL

Know Who I Am

José Teo Lorente Vallecillos

I'm a Head of AI & Data Engineering at Devera, where I lead the design and deployment of intelligent systems that drive real-world impact. With a strong foundation in computer science and a Master's in AI & Data Engineering (9.5/10 GPA), I specialize in building production-grade LLM agents, RAG architectures, and scalable data pipelines for domains like sustainability and healthcare.

Previously at Remote Care Partners, I developed AI-powered patient monitoring systems, achieving a 92% F1-score on hospitalization risk prediction and deploying quantized CNNs for real-time face authentication on edge devices. I thrive at the intersection of research and engineering — turning cutting-edge AI into robust, deployable solutions.

0+ Years in AI
0% F1-Score
0/10 MSc GPA
0% Top Percentile

Where I've Worked

Sep 2025 — Present

Head of AI & Data Engineering

Devera

  • Led redesign of core AI infrastructure, establishing scalable foundations for intelligent automation across the platform
  • Architected scalable data pipelines for carbon footprint estimation, enabling real-time sustainability analytics
  • Designed and deployed RAG systems for document understanding and knowledge retrieval at enterprise scale
  • Full lifecycle ownership from research and prototyping through production deployment and monitoring
LLM Agents RAG Data Pipelines Python AWS
Sep 2024 — Sep 2025

AI & ML Engineer

Remote Care Partners

  • AI-driven patient monitoring: built end-to-end systems for predictive healthcare analytics serving thousands of patients
  • Quantized CNN for face authentication: deployed lightweight deep learning models on edge devices for secure patient verification
  • Real-time biosignals integration: engineered pipelines processing continuous vital sign streams for anomaly detection
  • Achieved 92% F1-score on hospitalization risk prediction model, directly impacting clinical decision-making
PyTorch TensorFlow FastAPI Docker PostgreSQL
Mar 2024 — Aug 2024

AI & ML Intern

Remote Care Partners

  • End-to-end ML pipelines: designed and implemented complete machine learning workflows from data ingestion to model serving
  • Weather-aware health recommendations: built predictive models correlating environmental data with patient health outcomes
  • Achieved 45% latency reduction through model optimization and efficient inference pipeline architecture
Scikit-learn Pandas Python SQL REST APIs

What I Work With

PyTorch
PyTorch
TensorFlow
TensorFlow
LangChain
LangChain
RAG Systems
Scikit-learn
Scikit-learn
OpenCV
OpenCV
Hugging Face
Hugging Face
Deep Learning
NLP
Agent Orchestration
Python
Python
FastAPI
FastAPI
JavaScript
JavaScript
SQL
SQL
TypeScript
TypeScript
Flask
Flask
Docker
Docker
Kubernetes
Kubernetes
AWS
AWS
Azure
Azure
Git
Git
Linux
Linux
CI/CD
PostgreSQL
PostgreSQL
MongoDB
MongoDB
Pandas
Pandas
NumPy
NumPy
ETL/ELT
Data Modeling
Spanish
Native
English
C1
German
Basic
Chinese
HSK1

What I've Built

AGENTIC AI

GAIA Benchmark Agent — Top Performer

Certified Top Performer by Hugging Face with 90% on the GAIA benchmark. Agentic RAG system with multi-tool architecture featuring document retrieval, web search, code execution, and multi-step reasoning with robust guardrails.

LangChain OpenAI RAG Hugging Face
BACHELOR'S THESIS

AR System with AI Assistant

Bachelor's Thesis — 9.2/10. Real-time AR system integrating CNN-based facial authentication, NLP-powered assistant, and gesture control. Cost-effective VR-to-AR conversion via server-client streaming with multi-threaded GPU/CPU optimization.

OpenCV PyTorch Keras NLP
AI & GAME DEV

Maze Navigation AI

Intelligent agent system navigating dynamic mazes with wolf AI hunters. Features 4 terrain types, A* pathfinding with real-time decision-making, collision avoidance, and strategic evasion mechanics.

C++ A* AI Agents Pathfinding
3D GRAPHICS

3D Scene Renderer

Interactive 3D scene in C++ with OpenGL featuring handcrafted .ply models, objects of revolution, five simultaneous animations, multi-camera switching, and dynamic lighting with texture mapping.

OpenGL C++ GLSL 3D Math
ROBOTICS

Autonomous Sumo Robot — Finalist

University competition finalist. Autonomous robot with 3 ultrasonic sensors for opponent tracking, IR edge detection, dual Arduino boards, H-bridge motor control, and a custom black-tape boundary deception strategy.

Arduino C++ Sensors PCB
JAVA & OOP

Monopoly — MVC Architecture

Full Monopoly implementation in Java following the Model-View-Controller design pattern. Features a 2D graphical interface, OOP-driven game logic, player actions, and turn management with modular architecture.

Java MVC OOP 2D GUI
ALGORITHMS

N-Queens — 1st Place Winner

1st Place, University of Granada Programming Methodologies competition. Efficient C++ solution using backtracking and branch-and-bound algorithms for exhaustive enumeration of all valid queen placements on an NxN board.

C++ Backtracking Branch & Bound
AUTOMATA THEORY

Moore Machine Simulator

Automata and Formal Languages project implementing a Moore Machine with finite automata, stack machines, and Turing machine concepts. Explores language hierarchy relationships and computation models for language recognition.

Formal Languages Automata State Machines
REMOTE CARE PARTNERS

Patient Risk Prediction — 92% F1

AI-driven patient monitoring system analyzing real-time biosignals (ECG, SpO2, HRV) for hospitalization risk prediction. Achieved 92% F1-score with 30% false positive reduction, directly impacting clinical triage workflows.

PyTorch Time-Series FastAPI Healthcare AI
REMOTE CARE PARTNERS

Edge AI Face Authentication

Quantized CNN for secure facial authentication optimized for low-power NPU inference on healthcare edge devices. Lightweight deep learning model enabling real-time patient identity verification on constrained hardware.

CNN Quantization Edge AI NPU
REMOTE CARE PARTNERS

Real-Time Healthcare Pipeline

End-to-end ML pipeline for healthcare data: ingestion, feature engineering, training, and deployment. Reduced data processing latency by 45% through optimized ingestion and secure cloud storage for real-time analytics.

Data Pipelines AWS Docker PostgreSQL
DEVERA

Carbon Footprint Analytics Platform

Scalable data pipelines and AI services for automated product carbon footprint estimation. Architected the ingestion, orchestration, and analytics layers enabling real-time sustainability insights at enterprise scale.

Data Engineering Python AWS Sustainability
DEVERA

Enterprise RAG System

Production RAG system for internal knowledge retrieval and decision support. Designed document ingestion, vector search, and LLM orchestration layers deployed at enterprise scale with monitoring and optimization.

LLM RAG LangChain Vector DB

Achievements & Recognition

AI Agents Course

Excellence Top Performer

Hugging Face, 2025

90% on GAIA Benchmark

Data Engineering Bootcamp

Distinction

Dataexpert.io, 2024

Top 0.019% of 31,000 participants. Perfect Score.

Azure AI Fundamentals (AI-900)

Microsoft Certified

Microsoft Learn, 2024

Perfect Score

Fighting Robotics Finalist

Finalist

University of Granada, 2022

Autonomous sumo robot competition

N-Queens Challenge Winner

1st Place

University of Granada, 2020

Fastest optimized backtracking solution

Academic Background

MSc AI & Data Engineering

In Progress

ETSIIT — University of Granada

GPA: 9.5/10

Specialization in large language models, agentic systems, and scalable data architectures.

BSc Computer Science & Computer Engineering

ETSIIT — University of Granada

Ranked 52nd Worldwide in AI (U.S. News 2025)

Thesis: "Low-cost AR system with CV, NLP, and multimodal interaction" — 9.2/10

Get In Touch

I'm always interested in hearing about new opportunities, collaborations, or just having a conversation about AI and technology. Feel free to reach out through any of the channels below.