I build data systems that scale — reliable pipelines on Snowflake, clean dbt models, and semantic layers teams trust. Now extending that into applied GenAI: RAG systems, rigorously evaluated.
I'm a Data & Analytics Engineer working across the modern data stack — from raw ingestion to business-ready models. My day-to-day is designing dimensional data models, building tested dbt transformations on Snowflake, and shipping analytics that managers actually use.
My differentiator is connecting that foundation to trustworthy AI. My MSc thesis tackles uncertainty calibration in retrieval-augmented LLMs — whether a model's confidence can actually be trusted for risk-sensitive decisions, and how embedding quality shapes that. I care about AI that's grounded, measured, and honest about what it doesn't know — not just demoed.
Currently based in Germany and open to Data Engineering, Analytics Engineering, and AI Engineer roles across Europe.
Two roles across Germany and India — building data models, pipelines, and BI that drive decisions.
Selected work. Each project links to source code and documentation on GitHub.
A natural-language BI app for enterprise operations. Snowflake Cortex Analyst turns plain-English questions into governed SQL over a semantic layer; OpenAI turns results into executive summaries. Full transparency — every answer exposes the generated SQL, an interactive chart, and the raw data. Built with Streamlit on an end-to-end pipeline (S3 → Snowpipe → Snowflake → semantic view).
My MSc thesis on uncertainty calibration in retrieval-augmented LLMs — measuring whether a model's confidence can be trusted for risk-sensitive decisions, and how embedding quality shapes that reliability. A full pipeline over real financial filings: 5 embedding models compared, LLM-as-judge grading, and post-hoc calibration.
“Evaluating and Improving Uncertainty Calibration in Retrieval-Augmented LLMs: The Role of Embedding Quality in Risk-Sensitive Decision Making.”
A full data platform built the way a modern data team would: five business domains each owning their own data products, transformed with dbt (medallion), provisioned with Terraform, governed with tags/masking/row-access policies, enriched with Snowflake Cortex AI, and consumed through a Streamlit BI app with a grounded AI Copilot.
Grouped by discipline — depth where it counts, not a checklist.
I'm open to Data Engineering, Analytics Engineering, and AI Engineer roles across Europe. The fastest way to reach me is LinkedIn or email.