Leonardo – Python, LLM, AI agent development

A platform for real-time anomaly detection and monitoring of connected vehicle data, enabling automated identification of abnormal behavior and improving operational reliability through advanced data analysis.

• Architected and launched the Anomalous Stops detection platform, processing approximately 20M trips per day;
• Drove approximately $1M in annual cost savings through strategic platform migration initiatives;
• Released ML-enabled features to major fleet customers, averaging 5K+ vehicles.

Built a construction project management platform for tracking budgets, expenses, revenue, and project progress, providing end-to-end visibility into financial and operational performance throughout the project lifecycle.

• Developed the foundational system architecture and code patterns of a construction project management platform;
• Built the early stages of a successful platform for managing construction works and payroll;
• Designed frameworks and tooling that enabled the engineering team to develop 2x faster.

Built an AI-powered platform for extracting, classifying, and analyzing information from news articles, enabling automated processing of unstructured text to generate structured insights and support data-driven decision-making.

• Led the design and development of production-grade NLP and LLM systems for entity extraction from large-scale unstructured data sources;
• Architected end-to-end ML pipelines covering data collection, annotation, experimentation, fine-tuning, evaluation, deployment, and monitoring;
• Developed and productionized multilingual PyTorch models across 7 languages;
• Deployed GPU-backed BERT and coreference models into production, increasing extraction recall from 30% to 50%;
• Led technical decisions for ML platform initiatives across product, data, and engineering teams;
• Designed agentic AI platforms, including ground-truth generation, evaluation frameworks, fine-tuning, and release processes;
• Mentored engineers on ML engineering practices and scalable system design.

Built an AI and machine learning platform for automated extraction, classification, and analysis of information from news articles, transforming unstructured content into structured data for downstream processing and decision-making.

• Integrated machine learning models into high-throughput streaming services;
• Re-architected data streaming platforms, doubling extraction throughput;
• Led migration of ML-derived datasets into centralized cloud-based data platforms;
• Collaborated with ML engineers and data scientists to operationalize production models.