Skip to content
View moralesangel's full-sized avatar
👋
👋

Block or report moralesangel

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
moralesangel/README.md

Ángel Morales Romero

Data & AI Engineer | ML Production Systems Specialized in intelligent systems, production ML pipelines, and data quality automation. My work focuses on the intersection of anomaly detection, ML explainability, and scalable data engineering.

Education:

  • 🎓 MS in Robotics & Artificial Intelligence: Universidad de León (8.7/10, Thesis 9.2/10 honours) – Multimodal physiological signal processing for stress detection (EEG, ECG, electrodermal activity)
  • 🎓 BS in Computer Science: Universidad de Cádiz – Emergency vehicle detection in urban environments using Deep Learning

Current Role:

  • 💼 Data & AI Trainee @ HORSE (Renault Group) | Madrid, Spain
    Leading development of ADQG (Autonomous Data Quality Guardian) — a production ML anomaly detection pipeline on GCP/BigQuery
    • Deployed Isolation Forest, HDBSCAN, Fuzzy K-Means, and autoencoder-based anomaly detection models
    • Built full ipywidgets configuration UI with dynamic BigQuery column selection and cost estimation
    • Implemented SHAP explainability (beeswarm, waterfall, heatmap plots) for model interpretability
    • Designed Bronze/Silver/Gold medallion architecture with Dataform SQLX pipelines
    • Migrated workloads from Cloud Run Jobs to Cloud Batch with Artifact Registry containerization
    • Built Teams webhook alerting with Adaptive Cards; resolved GCP org policy violations for Vertex AI Workbench
    • Developed Gradio and ipywidgets interfaces for production ML configuration

Experience Highlights:

  • Production ML: End-to-end pipeline development (data ingestion → modeling → explainability → monitoring)
  • Data Engineering: BigQuery, Dataform, GCP infrastructure, Cloud Batch orchestration, IAM/service account management
  • ML Models: Isolation Forest, HDBSCAN, Fuzzy K-Means clustering, autoencoders with PCA preprocessing, SHAP interpretability
  • Infrastructure: GCP (Vertex AI, Cloud Run, Cloud Batch, Artifact Registry), Teams/Slack integrations, LZ1→LZ2 migration
  • Technical Depth: Sigmoid normalization for anomaly scores, cost optimization, data sufficiency validation

Research & Publications:

  • 🔬 Peer-reviewed publication: Neural Audio Classification for Emergency Vehicle Detection with Feature Compression via Convolutional Autoencoders
    ResearchGate

Technical Stack:

  • Languages: Python, SQL, C++, JavaScript, MATLAB
  • ML/Data: scikit-learn, Isolation Forest, HDBSCAN, SHAP, pandas, NumPy, PCA
  • Cloud & Infrastructure: GCP (BigQuery, Vertex AI, Cloud Batch, Cloud Run), Docker, Cloud Storage
  • Data Pipelines: Dataform (SQLX), dbt concepts, medallion architecture
  • Interfaces: ipywidgets, Gradio, Teams webhooks, Adaptive Cards
  • Other: ROS, React, Git

Languages:

  • 🌍 Spanish (Native)
  • 🌍 English (Professional Working Proficiency)

Location & Availability:

  • 📍 Madrid, Spain (Originally from Cádiz)
  • ✈️ Open to relocation / Remote roles
  • 🎯 Actively exploring: PhD programs (AI/ML research focus), production ML engineer roles, and AI research positions

Interests: Deep Learning, Computer Vision, LLMs, Anomaly Detection, ML Explainability, Reinforcement Learning, Generative AI

Looking to Collaborate On: Open Source ML & Robotics projects, production ML systems, anomaly detection applications

Current Goals: Building a generative AI research portfolio for PhD applications and senior ML engineer positions. Actively exploring: production LLM evaluation harnesses, percussive audio diffusion models, and sparse autoencoder analysis.


Connect:

Popular repositories Loading

  1. lettersnumbers-classifier lettersnumbers-classifier Public

    Letters and numbers classifier for chess games using computer vision and neural networks.

    MATLAB

  2. fire-detection fire-detection Public

    This project will try to segment forest images taken by a drone, classifying each pixel by either forest or fire (segmentation technique)

    MATLAB

  3. coins-detection coins-detection Public

    This project will try to detect every type of coin within an image using Computer Vision techniques.

    MATLAB

  4. slime-mold-simulation slime-mold-simulation Public

    This project implements a simulation of the slime mold behavior using p5.js, inspired by the movement and pattern formation of real slime molds. The slime molds are represented as particles that le…

    JavaScript

  5. emergency-vehicle-detection emergency-vehicle-detection Public

    A compact pipeline for detecting emergency vehicles from audio using deep learning. Includes data download and preprocessing, MFCC/LFCC/Chroma feature extraction, model training (autoencoders for d…

    Jupyter Notebook

  6. BPE-tokenizer BPE-tokenizer Public

    A clean, educational implementation of the Byte Pair Encoding algorithm used in modern language models like GPT.

    Jupyter Notebook