Hi there, I'm Lucas Ercolano 👋 AI Consultant | AI Engineering Student @ UdeSA
I am passionate about artificial intelligence and machine learning, fascinated by how these technologies are transforming the world, and driven to be part of that revolution. I am resourceful, proactive, and thrive on solving complex challenges—which is why I chose to specialize in AI/ML. Beyond code, I am a creative thinker who loves exploring original ideas, working collaboratively, and bringing high energy to diverse teams.
🚀 Featured Projects NVIDIA Alpamayo Adaptation for Autonomous Driving (Feb 2026 – Present) Adapting NVIDIA's reasoning VLA model (Alpamayo) to a private, real-world multi-camera vehicle dataset recorded in Argentina. This involves reconfiguring the original 4-camera pipeline to a custom 5-camera rig from first principles, without ground-truth labels. The focus is on qualitative, behavioral evaluation of the model's chain-of-causation reasoning traces in chaotic, unstructured traffic environments. Currently integrating Alpamayo 1.5 features like RL post-training and navigation guidance.
Privacy-Preserving Perception Pipeline (Jan 2026 – Present) Built a privacy-aware perception pipeline for a real autonomous vehicle with a stereo camera rig. Fine-tuned Meta's EgoBlur Gen2 on a custom ~11k image dataset to automatically detect and blur license plates and faces. The model acts as a teacher in a knowledge distillation pipeline to train a highly efficient YOLOv11s student model.
OpenClaw AI Agents: Rook & mailbot (Current) Developing autonomous AI agent systems using the OpenClaw framework. This involves configuring complex agent behaviors, managing environment interactions, and optimizing the underlying architecture for robust, task-oriented execution.
Self-Supervised Lane Detection (Jul 2025 – Nov 2025) Developed a robust lane detection system using a custom, unlabeled dataset. Built a teacher-student knowledge distillation pipeline where a high-capacity teacher model generates pseudo-labels to guide a lightweight student model, achieving real-time inference speeds for autonomous driving.
Gomoku AI Agent (Jul 2024 – Nov 2024) Created a superhuman AI agent for the strategic board game Gomoku. Leveraged classical AI algorithms and advanced optimization techniques to mimic and exceed human decision-making. Summarized in a poster and presented at the Scientific Symposium on Artificial Intelligence and Applications (SCIAA).
Link'Em (Mar 2024 – Apr 2024) A B2B marketplace platform built with Node.js, HTML, and CSS, designed to connect new entrepreneurs with wholesale suppliers.
💼 Experience AI Consultant @ H5K5 Leak Detection (Apr 2025 – Present)
AI & Digital Transformation Specialist @ Municipalidad de San Nicolás de los Arroyos (Dec 2024 – Mar 2025)
Designed end-to-end ML, NLP, and Computer Vision solutions to modernize public administration, including large-scale conversational systems and real-time traffic analysis.
University Assistant Professor @ Universidad de San Andrés (Feb 2024 – Jun 2024)
Software Developer & IT Support @ Don Gino Repuestos (Jul 2022 – Feb 2023)
IT Technician @ Accounting Firm (May 2022 – Jul 2022)
🎓 Education & Certifications Universidad de San Andrés: Artificial Intelligence Engineering (2023 – 2028)
Technical AI Safety Course: BlueDot Impact (Mar 2026)
IELTS Academic (Advanced English): University of Cambridge (Dec 2025)
Emprelatam Startup Scholarship: ITBA Computer Society Hackathon Winning Project (Apr 2024)
Itba Game Jam Champion: ITBA Universidad (Nov 2023)
Argentine Mathematical Olympiad Champion (Nov 2017)
🌍 Languages Spanish: Native/Bilingual
English: Full Professional Proficiency (Advanced)
Italian: Limited Working Proficiency