AI & Machine Learning

Experience with AI & Machine Learning

Artificial intelligence and machine learning are at the core of OpenMyPro's competitive advantage — the AI-powered matching algorithm that connects patients with the right healthcare provider in under two seconds is the technology that enables the platform's signature 33-second booking experience. Pablo Diaz built this AI system from the ground up using Python, scikit-learn, and custom feature engineering tailored to the unique requirements of healthcare provider matching. The matching algorithm processes multiple dimensions simultaneously: patient stated preferences (specialty type, gender preference, language, availability), geographic proximity using PostGIS spatial queries, provider credentials and verification status, historical booking patterns (providers who retain patients have higher match scores), insurance compatibility when applicable, pricing transparency for cash-pay patients, and real-time availability from provider calendars. These features are weighted and combined into a composite relevance score using a gradient boosting model trained on historical booking and retention data — providers that patients book with and return to receive progressively higher match scores for similar patient profiles. The training pipeline runs in a containerized Python environment: data extraction from Supabase, feature engineering with pandas, model training with scikit-learn, hyperparameter tuning with cross-validation, and model deployment through a FastAPI inference endpoint that the Next.js frontend calls. The model is retrained weekly on new booking data, ensuring the matching improves over time as the platform accumulates more signal on which patient-provider pairings lead to successful outcomes. Beyond the matching algorithm, Pablo has implemented AI-powered features including automated provider specialty classification from profile descriptions using NLP, smart search with semantic understanding (searching 'stress help' returns therapists even though the term 'stress' may not appear in their profile), and anomaly detection for identifying unusual booking patterns that may indicate platform abuse. His AI/ML expertise is practical and product-focused — every model serves a specific user need, and every feature is evaluated against the metric that matters most: does it reduce the time from search to booked appointment?

Frequently Asked Questions