Loading headlines…
Niher Halder

Projects

A focused set of projects demonstrating capability in AI/ML, programming, and automation, developed with an engineering mindset. I emphasize clear documentation, validation, and evidence, and I am progressively aligning this work toward OT/ICS-focused anomaly detection and monitoring for power and industrial systems.

Quality > Quantity

Each project includes metrics, decisions, and documentation — not just code.

Documentation Metrics Reproducible

Engineering mindset

I treat ML work like engineering: constraints, validation, and evidence.

Validation Risk-aware Practical

OT/ICS direction

Next: anomaly detection / monitoring projects aligned to gas & power OT systems.

Time-series Anomaly detection OT/ICS

Featured

🏠 House Price Prediction (XGBoost)

End-to-end ML workflow: feature engineering, cross-validation, tuning, and deployment mindset.

XGBoost Feature engineering Cross-validation
  • CV RMSE: 0.1229 (log-space)
  • Kaggle Score: 0.12826
  • Status: ✅ Completed
Read Case Study GitHub

🧠 Facial Keypoints Detection (CNN)

Deep learning practice: model building, training stability, evaluation discipline, and error analysis.

CNN TensorFlow/Keras Computer vision
  • Val RMSE: 0.0230
  • Val MAE: 0.0163
  • Status: ✅ Completed
Read Case Study GitHub

Next Project (OT/ICS — In Progress)

Planned: OT/ICS anomaly detection (time-series telemetry + operationally meaningful alerts). Will include dataset choice/simulation, baseline method, model comparison, false-positive control strategy, and a security architecture diagram.

Time-series Explainable alerts Gas & Power OT/ICS