About Me

Hallo!

I am a machine learning engineer with more than 10 years of experience turning complex data challenges into production-ready ML systems. I love building scalable systems powered by AI/ML. I follow a three-step framework: Feasibility, Prototype, and Production.

I specialize in building AI/ML systems at scale that deliver real business impact. I speak english, spanish and basic german.

Key Projects & Impact

I’ve worked on projects involving mainly text and tabular data using deep learning and general ML algorithms:

Agentic RAG Chatbot

Built and deployed a complex agentic RAG chatbot with high accuracy of responses and handling up to 100 RPS.

LLMs RAG Agentic AI

LLM Fine-Tuning

Supervised fine-tuning of LLMs to create clones of various characters with distinct personalities.

LLMs Fine-Tuning Deep Learning

Ads Ranking Systems

Built production ranking models to rank restaurants based on click probability, improving user engagement and conversion rates.

Python ML A/B Testing

Search & Relevancy

Developed and deployed search models to enhance relevancy in search results, significantly improving user satisfaction metrics.

NLP Information Retrieval Python

Recommendation Systems

Built recommendation models improving user discovery of products on platforms, driving increased engagement and sales.

Collaborative Filtering Deep Learning Production ML

Text Duplication System

Created a data product using open source technologies to efficiently identify duplicate text at scale.

NLP Microservices SQLite FTS

Cost Optimization

Migrated data science project from Spark to Python in-memory computation using Polars, saving hundreds of dollars weekly in cloud costs.

Polars Performance Cost Reduction

Open Source Contributions

superml - R Package

I am the author and maintainer of superml, an R package that brings scikit-learn-like functionality to R users. The package helps data scientists build ML models in R with an intuitive API similar to Python’s scikit-learn.

Key Features:

Scikit-learn inspired API design
Support for common ML algorithms
Easy-to-use interface for R users
Active maintenance and community support

Talks & Presentations

I enjoy sharing my knowledge through talks and presentations. Here are some of my featured talks:

Better Search Relevance with XGBoost

Walk though building a learning to rank ML model and API to provide relevant search result

Pycon Berlin • 2024

Watch on YouTube

Learning to Rank with XGBoost

Using XGBoost to build a LTR model and walk through the service architecture

Berlin Buzzwords • 2024