10+ years across ML, data, backend, and product. Currently at Lumber building AI systems for document processing and automation. Previously shipping RAG systems and AI agents for Fortune 500 customers.
Currently at Lumber as AI Engineer, building AI systems for document extraction and compliance automation. Previously at QuarkAI, where I worked on RAG systems and AI agents for Fortune 500 customers, and co-authored a paper submitted to the NAACL 2025 Industry Track.
Before that: Radium Rocket (Data & Backend), Ternium (Data Analyst Team Lead at Latin America's largest steel company).
Co-founded an Electronics & IoT startup that reached 25+ multinationals, 10+ research institutes, and 15+ government institutions. Led a cross-functional team through two successful launches. Distinguished by the Chamber of Deputies.
During COVID-19, built covidargentina.com.ar (+1000 monthly users), a data visualization dashboard in collaboration with NECSI, MIT, Harvard, and CONICET. Appointed Latin American Representative Data Scientist by EndCoronavirus. Distinguished by the Municipal Council of Rosario.
I work well in ambiguity, lead teams when needed, and care about the product, not just the code.
Building AI systems for document processing and automation. Work spans OCR/LLM pipelines, extraction systems, agentic workflows, and production ML infrastructure.
Contributed to QuarkAI's flagship RAG-based SaaS application QuarkGPT for customer support, focusing on shortlisting, system optimization, and designing prompt strategies. Led production releases to Fortune 500 companies.
Developed support tickets summarization and tagging, generating synthetic representations and categories for customers' data using LLMs like OpenAI's GPT and Anthropic. Designed the prompt strategy.
Developed a "proactive" AI agent that autonomously monitors support tickets, using LLMs such as Claude and GPT and information retrieval tools to assess when to take action, like notifying teams or issuing alerts.
Researched and implemented accuracy enhancements in vector-based semantic search by leveraging LLMs to extract metadata alongside embeddings. Co-authored paper submitted to NAACL 2025 Industry Track.
Designed and trained computer vision models in PyTorch and TensorFlow to recognize documents' layouts. Optimized with ONNX.
Implemented expert knowledge capture from chat logs, involving threads identification and question-answer extraction.
Orchestrated data ingestion pipelines (connectors, ingestion, indexing) for multiple customers using Apache Airflow, AWS Services (S3, SQS, Cloudwatch) and Apache Solr.
Designed and developed data ingestion engines for PDF and XLSX files that transformed complex hierarchical data into semantically segmented entries.
Implemented a web crawler in Java featuring a novel fetcher strategy that combines web drivers and HTTP requests. Designed and implemented incremental ingestion systems to periodically update models by monitoring clients' API endpoints.
Trusted by more than 25+ multinational companies, 10+ research institutes and universities, and 15+ government institutions.
Led a cross-functional team of 10+ people (engineers, designers, and marketers) to two successful product launches: portable and IoT CO2 monitors to measure indoor air quality in the context of the COVID-19 pandemic. Shipped MVP in 30 days, first full-featured product in 90 days.
Designed and implemented an IoT centralized system (using MQTT and MERN stack for the platform) for ESP32 processor.
Web apps development and maintenance for U.S. startups and enterprises through outsourcers.
Engineered ETL pipelines with Apache Airflow, AWS Lambda, AWS S3, AWS EC2, Google Compute Engine (GCE), Python (Pandas, Numpy).
Implemented crawlers, scrapers, and other automations using Selenium, Scrapy-Splash.
Python backend: Flask, Gunicorn, Nginx, combined with PostgreSQL, MongoDB, and Solr.
Leader of contractor management. Training and support for managers and peers in Argentina, Brazil, Mexico, Colombia, Guatemala, USA.
Implemented a SQL database for 20,000 contractors across 400 companies, managed the annual economic budget of the labor force.
Led a cross-functional team in preparing and presenting monthly reports to executives.
Developed automation systems for personnel management, resulting in a reduction of 20% in manual effort and increased accuracy.
Optimization of labor productivity. Reporting, Microsoft Power BI dashboards. IT projects coordination. Benchmarking between production units, plants, processes, lines, and other companies to optimize industrial processes efficiency.
Standardization and definition of procedural processes of the area: objectives, scope, inputs and outputs. Definition and monitoring of KPIs.
Designed and developed an interactive data visualization dashboard to track the COVID-19 pandemic (+1000 users monthly), collaborating with global and local research institutions: New England Complex Systems Institute, MIT, Harvard, Brandeis, CONICET. Implemented a daily automated pipeline to ensure consistently updated data from multiple sources, with district granularity level coverage.
Researched and implemented accuracy enhancements in vector-based semantic search by leveraging LLMs to extract metadata alongside embeddings. Co-authored paper submitted to NAACL 2025 Industry Track.
End-to-end IoT platform for indoor air quality monitoring during COVID-19. MQTT-based sensor network with MERN stack web platform for ESP32 processors. Deployed across schools, hospitals, and corporate offices.
Developed an app to generate PowerPoint presentations from videos that outputted editable .ppt presentations using Tesseract.
RAG accuracy enhancements in vector-based semantic search by leveraging LLMs to extract metadata alongside embeddings.
Recognized for contribution to public health through IoT air quality monitoring during the COVID-19 pandemic.
Appointed by the international volunteer coalition based at the New England Complex Systems Institute.