Ivo Tonioni
GitHubLinkedInEmail

Ivo Tonioni

Quantitative ML Engineer

Energy Markets · Time-Series Forecasting · Applied AI

Building forecasting and decision systems for energy markets.

Machine learning, quantitative research, and power-systems context for validated forecasts and risk-aware decisions.

Based in Copenhagen, Denmark · Open to ML, quant research, and energy forecasting opportunities

View flagship projectGitHubLinkedInDownload CV
DK2 / 24Hwalk-forward
Forecast versus actual electricity price chartA static abstract line chart comparing an actual time series against a forecast.
Flagship project

DK2 Electricity Price Forecasting & Trading Research

24-hour-ahead electricity price forecasting for the Danish DK2 bidding zone.

Data

Prices, weather, load, wind, solar, renewables, and net load.

Modeling

Baselines, XGBoost, and PyTorch LSTM benchmark.

Validation

Leakage-aware temporal splits and expanding-window walk-forward evaluation.

Decision layer

Forecast-to-signal logic, PnL, drawdown, VaR, and loss-limit analysis.

Engineering

Modular Python for features, training, validation, backtesting, and risk checks.

Backtest disclaimer

Simplified research backtest; excludes full frictions, liquidity, transaction costs, imbalance costs, and operational constraints.

View GitHub repositoryRead methodology
Research workflow

From data to risk-aware decisions

  1. 01

    Market & weather data

    Align price, load, generation, and weather signals.

  2. 02

    Feature engineering

    Build lags, calendar features, renewables, and net load.

  3. 03

    Forecasting models

    Compare baselines, XGBoost, and sequence models.

  4. 04

    Walk-forward validation

    Evaluate with expanding temporal windows.

  5. 05

    Backtesting

    Translate forecasts into simple signals.

  6. 06

    Risk controls

    Track drawdown, VaR, loss limits, and failure modes.

Selected work

Focused forecasting and infrastructure work

Cryptocurrency Price Forecasting with Autoformer

MSc thesis on transformer-based forecasting for volatile crypto time series, benchmarked against LSTM.

  • Python
  • PyTorch
  • Autoformer
  • LSTM
  • Time Series
  • Financial Data

ML Infrastructure & Reproducible Deployments

Local AI and Streamlit deployments using Docker, k3s, GitHub Actions, and Raspberry Pi infrastructure.

  • Docker
  • Linux
  • k3s
  • Streamlit
  • GitHub Actions
  • MLOps
Domain experience

Power systems context

Denmark

Bravida / Energinet Projects — Denmark

HV Substation, Protection & Control Interface

  • Coordination with Ramboll and Energinet on power infrastructure.
  • Commissioning readiness, P&C interfaces, wiring, loop checks, and documentation.
  • Grid assets, operational constraints, and reliability requirements.
Technical stack

Tools for forecasting, validation, and delivery

Machine Learning

  • Python
  • pandas
  • NumPy
  • scikit-learn
  • XGBoost
  • PyTorch

Quant Research

  • Time Series
  • Feature Engineering
  • Walk-Forward Validation
  • Backtesting
  • PnL
  • Drawdown
  • VaR

Energy Systems

  • Electricity Markets
  • Load
  • Wind
  • Solar
  • Net Load
  • HV/MV Infrastructure
  • Protection & Control

Engineering

  • Git/GitHub
  • Docker
  • Streamlit
  • Linux
  • Jupyter
  • CI/CD

Contact

Let’s build better decision systems for energy markets.

Open to ML engineering, quant research, and energy forecasting roles.

Email meLinkedInGitHub