Jose – Python, AWS, AI agent development

A Databricks-based machine learning platform powering data-driven products across 50+ countries. Automates end-to-end ML workflows—from data ingestion and feature engineering to model training, hyperparameter optimization, and inference—enabling data science teams to deploy real-time insights for consumer apps, supply chain optimization, and product development at global scale.

• Developed and maintained large-scale ML pipelines for training, inference, and hyperparameter tuning.
• Built and optimized Databricks jobs for high-volume, multi-market datasets.
• Automated end-to-end ML workflows, including data ingestion, feature engineering, training, and evaluation.
• Designed reusable and reproducible pipelines for training and deployment.
• Improved performance and cost-efficiency through cluster optimization and job orchestration.
• Implemented shared libraries and templates to standardize ML development.
• Supported migration and stabilization of workflows across environments.
• Collaborated with global teams to ensure scalability and consistency.

A computer vision solution designed to measure the effectiveness of cosmetic products over time through longitudinal analysis of facial images. The platform includes longitudinal analysis of treatment effectiveness, multidimensional color space computations, skin issues (wrinkles, oiliness, spots) estimation using 3D surface modeling and AI models, customer segmentation models, and YOLO-based object detection and segmentation models.

• Built computer vision models to quantify wrinkles, spots, oiliness, skin tone, and other dermal features.
• Developed GenAI-based image simulation features for wrinkle and blemish reduction.
• Implemented an LLM-powered recommendation system for personalized cosmetic suggestions.
• Created classifiers for skin tone, skin type, oiliness, and color matching.
• Designed clustering models for customer segmentation and targeting.
• Built batch and real-time ML pipelines for inference and reporting.
• Developed and maintained AWS-based ML infrastructure supporting multiple AI products.
• Automated manual and lab workflows, generating over US$3M in savings in the first two years.

A large-scale legal document classification system using NLP techniques and transformer-based models. The solution processes high volumes of legal data and delivers accurate classification results in a production Databricks environment.

• Improved model accuracy from ~80% to above 90%.
• Applied BERT and classical ML techniques for text classification.
• Built scalable preprocessing pipelines for large legal document datasets.
• Developed feature engineering workflows.
• Automated training and inference pipelines in Databricks.
• Delivered a production-ready NLP system for enterprise use.

A data automation and web scraping startup that turns scattered online and internal data into clean, actionable insights for decision‑making. It builds custom data-extraction robots, integrates marketing and financial sources (ad platforms, bank data, CRMs), and enriches lead databases so companies can monitor competitors, optimize campaigns, and streamline reporting directly into tools like spreadsheets and internal systems.

• Built deep learning models (TensorFlow, PyTorch, Keras) for mask and helmet detection.
• Developed a real-time mask detection system integrated with Android devices for access control (e.g., gym turnstiles during COVID-19).
• Implemented computer vision solutions for workplace safety compliance (PPE detection).
• Developed OCR pipelines for document analysis using OpenCV and Tesseract.
• Performed image preprocessing and feature extraction for improved model performance.
• Exposed ML models via APIs using Docker.
• Integrated ML models into Android applications using Java and Android Studio.

A real estate intelligence platform designed to support multimillion-dollar investment decisions for large construction companies. The product delivered data-driven insights about geographic regions, replacing subjective analysis with objective metrics for pricing, customer targeting, and project feasibility.

• Designed and implemented core ML and data processing components for the MVP.
• Developed data pipelines and backend logic to support product functionality.
• Collaborated closely with product and business stakeholders to prioritize features.
• Built scalable and production-ready components under startup constraints.
• Contributed to the product launch and early traction.
• Supported the startup in reaching break-even through efficient MVP delivery.