Self-directed project
Energy and Renewable: Machine Learning for the Power Grid
Junior Machine Learning Engineer · AI / ML · August 2026
A ProoV case study · educational project, not employment
Built a turbine energy-yield predictor end to end using leakage-free sensor inputs, a time-based train/test split, and a RandomForest model that outperformed a LinearRegression baseline. Reported honest held-out performance (MAE, RMSE, R²), read residual behavior, and documented intended use, exclusions, limitations, and monitoring in a concise model card.
Graded against
- Build and validate the model correctly35%
- Optimize and evaluate honestly25%
- Write the Model Card (clarity + honesty)25%
- Explain it to a non-expert15%
Passed · pass mark 60/100
What stood out6
- Correctly excluded CO and NOx as downstream leakage features and kept the eight legitimate ambient/turbine sensors.
- Used an earlier-period train / later-period test split instead of a random shuffle, preserving time order.
- Compared a LinearRegression baseline against a tuned RandomForest and reported the improvement on held-out data.
- Included a concrete monitoring plan with distribution checks and a retrain threshold tied to error drift.
- Leakage-Free Feature Selection
- Honest Held-Out Evaluation
The work I submitted18 tasks
10 tasks · 5.1k characters · Python · 8 written answers · 251 words
- Problem Frame226 characters
- Leakage Flags113 characters
- Train Validate BaselinePython · 893 characters
- Train Validate Forest775 characters
- Train Validate124 characters
- Tune Evaluate SetupPython · 1.5k characters
- Tune Evaluate91 characters
- Residual Read224 characters
- Drift Monitor91 characters
- Model Card1.1k characters
- Problem Frame17 words
- Leakage Flags1 word
- Train Validate Baseline61 words
- Train Validate Forest44 words
- Train Validate1 word
- Tune Evaluate Setup100 words
- Tune Evaluate1 word
- Residual Read26 words
Summarised on purpose — the submitted code and writing stay private so this page cannot be reused as an answer key. The full submission sits behind the verified certificate.