Selected projects and case studies from the fields of Data Science, Machine Learning, and data analytics. Each project showcases a specific problem, the solution used, and measurable results.
Results in Numbers
| Metric | Value |
|---|---|
| Completed projects | dozens |
| Production ML models (Unified Pipeline) | ~100 |
| Years of DS/ML experience | 7+ |
| Average model accuracy improvement | 15% |
| Model deployment time reduction | up to 80% |
Selected Projects
⚙️ Unified Pipeline
Problem: Fragmented ML processes, inconsistent model quality, and long deployment times (10+ days).
Solution: Design and implementation of a unified ML pipeline on the Databricks platform with automated feature engineering, model registry, and A/B testing.
Results:
| Metric | Before | After | Improvement |
|---|---|---|---|
| Model deployment time | 10 days | 2 days | −8 days |
| Model accuracy | baseline | +15% | +15% |
| Pipeline runtime | 4 hours | 2.4 hours | −40% |
Technologies: Python, PySpark, Databricks, MLflow, Optuna, Docker
🏠 PENB Energy Label Approximation
Problem: A formal energy audit is often too slow and too expensive for an initial decision, while the user still needs a fast signal grounded in real operational data.
Solution: A public web application that combines input validation, weather data, a simplified RC model, and an interpretable result for screening an apartment before a deeper assessment.
Results:
| Metric | Status |
|---|---|
| Public application | yes |
| Language versions | 2 |
| Computation modes | 3 |
| Result export | HTML report |
Technologies: Python, Streamlit, Docker, Pandas, Plotly
🤖 Open Source Projects
- MCP Prompt Broker – routing of instruction profiles for AI agents based on prompt type
- WireGuard in Docker – VPN + Nginx reverse proxy architecture for secure service publishing
Areas of Expertise
- Banking and finance: Propensity models, CRM optimization, NBO, dynamic pricing, product analytics
- MLOps: ML pipeline automation, deployment, monitoring, model registry (MLflow)
- Big Data: PySpark, Databricks, Spark, Hadoop, Hortonworks, Oracle DBMS, distributed computing
- BI & Reporting: Power BI, MS Excel, PL/SQL analytics, data transfers and integrations
- Econometrics: Macroeconomic analyses, forecasting, time series
- AI Engineering: MCP, LLM integration, RAG pipeline, AI implementation in corporate environments
Interested in What I Can Do for You?
If you are working on production ML deployment, scoring, or a realistic AI use case for your company, we can start with a free 20-minute introductory consultation and quickly assess where collaboration creates the most value.