Featured Projects
Selected work spanning AI research, production systems, and infrastructure
Master's thesis research at FCEyN — University of Buenos Aires. Implemented a full debate protocol simulation using PyTorch — including computer vision models, turn-based strategies, and agent asymmetry. Empirically demonstrated that debate improves accuracy of limited judges in AI decision-making.
Semantic Search RAG System
Production-grade agentic RAG system using Gemini LLM and Google embeddings for natural language queries over a 544-product catalog. Vector storage with pgvector, deployed on AWS with Flask APIs.
Melanoma Detection Model
PyTorch-based computer vision system for melanoma detection achieving 80% recall. Includes CNN architectures, OpenCV preprocessing pipelines, and comprehensive model evaluation.
Personal Cloud Infrastructure
Self-designed cloud infrastructure for deploying ML applications, research demos, and web services. Docker containerization with reverse proxy routing, SSL certificates, domain management, and uptime monitoring.
Want to see more of my work?
View all on GitHubExperience & Education
From academic research to production AI at global scale
Professional Experience
AI Engineer (Sr. Analyst)
MERCADO LIBRE
Leading AI initiatives including LLM-powered compliance automation and a strategic product collaboration with Google, serving 200M+ users across Latin America.
AI Engineer & Team Lead
NETV S.A.
Lead of AI & Data team. Built production semantic search for 544+ products. Optimized SQL pipelines 97% faster (40 min to under 60s).
Data Engineer & ML Consultant
LAS MERCEDES
Automated reconciliation for 500+ monthly POS transactions, reducing manual processing by 85%. Applied ML forecasting to reduce stockouts by 25%.
Education & Certifications
Master's in Data Science (Artificial Intelligence)
University of Buenos Aires
Bachelor's in Data Science (Mathematics + Computing + Statistics)
University of Buenos Aires
English Proficiency (FCE B2)
Cambridge English
About Me
I'm an AI/ML Engineer from Buenos Aires with 4+ years turning research into production systems. Currently leading AI initiatives at MercadoLibre (MELI), I specialize in LLM-powered automation, agentic architectures, and multimodal AI pipelines deployed at global scale.
I hold a Master's in Data Science from the University of Buenos Aires, where I also research AI Safety at FAIR-IALAB — investigating how debate mechanisms can make AI systems more trustworthy. Outside work, I run my own cloud infrastructure for deploying ML applications and experiments.
I enjoy collaborating on challenging AI problems — from research to production.
Technical Skills
Core technologies I use to build and deploy AI systems at scale
AI/ML & LLMs
Cloud & Infrastructure
Data & Backend
Teamwork & Communication
Let's Connect
I enjoy connecting with people working on interesting problems.
Interested in working together?
My expertise spans LLM systems, RAG architectures, agentic AI, and end-to-end ML engineering at scale.
© 2026 Joaquin Salvador Machulsky