ENERGY DATA ANALYST — German power markets · BESS · ML forecasting
Munich, Germany · connectashish28@gmail.com · LinkedIn
I work where battery hardware meets the German power market. At Entrix, I build production analytics for grid-scale BESS — multi-market backtesting, customer-facing dashboards, commercial enablement. Master's in Energy Science from Ulm University, with prior R&D at DENSO and ML research at TH Ulm.
I read battery operations as well as I read market data.
- Grid-scale BESS analytics — multi-market optimization (FCR, aFRR, EPEX DA/ID), revenue backtesting, customer-facing analytics
- Production data tooling — Grafana, Preset/Superset, Streamlit on AWS (ECS, S3, Athena)
- ML for energy — load and PV forecasting (seq2seq LSTM with quantile heads), CVAE-based synthetic load profiles, SOC estimation on real BMS data
| Project | What it does | Stack |
|---|---|---|
| German Load Forecast ↗ | Beats Germany's official TSO day-ahead load forecast by 21.5% (up to +39% on extreme days). seq2seq LSTM with P10/P50/P90 quantile heads on public SMARD + weather data; leakage-safe pipeline, daily refresh. Live demo ↗ | TensorFlow Streamlit SMARD GH Actions |
| Redispatch Dashboard ↗ | Live dashboard over 2 years of German DSO redispatch activity across 175 Schleswig-Holstein substations. 15-min parquet pipeline, daily map + 90-day congestion KPIs. Live demo ↗ | Streamlit Plotly parquet SMARD |
| BMS SOC Estimation ↗ | Cycle detection and coulombic-efficiency drift correction on real BMS telemetry from a 14 kWh LFP pack, validated against BMS-reported SOC. | Python InfluxDB |
LANGUAGES Python · SQL · MATLAB
ML & DATA pandas · NumPy · scikit-learn · TensorFlow · PuLP
CLOUD AWS (ECS, S3, Athena) · Docker · InfluxDB · DynamoDB
BI & VIZ Grafana · Preset/Superset · Streamlit · Plotly
DOMAIN EPEX (DA/ID, FCR, aFRR) · Redispatch · BESS · SOC/C-rate
GET IN TOUCH