I’m a full-stack Data Scientist and Engineer with a background that bridges applied AI and business strategy.
I currently work for a large financial group’s AI and Data competence center, helping modernize the company’s capabilities and translate AI and data-related business needs into operational, ROI-driven solutions.
My approach is simple: experiment quickly, validate assumptions early, and build only what delivers measurable business impact.
I thrive where technical depth meets strategic thinking, and where the goal isn’t just to deploy AI, but to make it matter.
PyTorch, Deep Learning, Time-Series Forecasting, LLMs, RAG Systems, Computer Vision.
Databricks, Azure, PySpark, SQL, MongoDB, FastAPI, Docker.
Explores the application of deep learning techniques to predict the cross-section of stock returns. The focus is on developing and automatically tuning deep neural network models (AutoML) to enhance forecast accuracy. This study leverages advanced machine learning methodologies to assess the performance of semi-automated methodologies in asset pricing.
A full-stack Retrieval-Augmented Generation (RAG) application that turns Obsidian notes into a queryable intelligence system. It empowers users to have natural language conversations with their personal knowledge base. Features a robust pipeline for uploading vaults, automated metadata extraction, and intelligent duplicate detection.
Developed a large-scale forecasting system capable of generating coherent probabilistic predictions across
multiple hierarchical levels. The solution makes use of both Machine Learning and Deep Learning models
to extract maximum predictive performance.
The probabilistic reconciliation ensures that forecasts remain consistent across all aggregation layers.
The models and reconciliation methods are selected through an extensive backtesting and optimization
framework, evaluating performance, calibration, and hierarchical coherence under different model
classes and uncertainty quantification techniques.
A lightweight web application capable of transcribing YouTube videos, podcasts, and uploaded audio files. Leverages OpenAI’s Whisper model for speech-to-text with support for both local inference and API-based processing. Designed to demonstrate practical integration of AI models in web applications with a user-friendly interface.
Bridging the gap between Jupyter notebooks and production.
Proficient in wrapping models in APIs (FastAPI), containerization (Docker), and cloud deployment to ensure models provide value in real-time applications.
Experienced fine-tuning open-source models LLMs and SLMs and architecting privacy-first, permissioned, RAG pipelines and multi-agent systems.
Specialized in probabilistic forecasting and hierarchical reconciliation for high volumes of time series.
Proficient in using Statistical models, Machine Learning and Deep Learning to predict complex time series.
Whether you have a question about a project, want to discuss a potential collaboration, or just want to talk about the latest in AI, I'd love to hear from you.
Or connect with me on social media: