Production-oriented ML engineering workspace. target repo ini jadi kumpulan project end-to-end untuk pipeline training, evaluation, artifact logging, model comparison, dan production readiness.
| Project | Description | Status |
|---|---|---|
| fraud-detection-pipeline | Fraud detection pipeline dengan chronological split, leakage checks, feature engineering, imbalance handling, threshold tuning, artifact logging, latency metric, dan multi-model tuning (LightGBM, Random Forest, XGBoost). | Active |
| indonesian-banking-asr | Indonesian banking ASR dengan synthetic dataset + Gemini/9Router, entity-aware WER, augmentation, dan Whisper/LoRA eksperimen. | Active (WIP) |
| kaggle-comp/house-prices-advanced-regression-techniques | Kaggle House Prices regression framework dengan fitur engineering, multiple models, dan artifact/submission logging. | Inactive |
| kaggle-comp/neurogolf-2026 | Kaggle ARC-like ARC task workspace untuk ONNX model constraints dan submission pipeline. | Inactive |
| gameboard-bot | Self-play loop + training pipeline untuk bot Wall Chess/Quoridor, fokus distillation ke evaluator cepat untuk Rust/Browser. Live demo (main langsung di browser via Rust/WASM). | Active (R&D) |
| webcam-rts-effect | Webcam realtime effects dengan segmentation, classification, hand tracking, web editor, serta tooling runtime untuk holdout recording, evaluasi model, benchmark, dan diagnostics. | Active (Runtime Evaluation) |
| ml-production-ecosystem | Toolkit scaffold dan production-playground untuk generate project ML lifecycle starter (serving, monitoring, retraining, rollback). | Active / Foundation |
Daftar ini otomatis direfresh berdasarkan isi katalog project dan README masing-masing.
Personal use.