- 🚀 0 → 1 builder: research prototype to deployed product, API included.
- 🧪 Diagnostics before hype — validated signals, honest metrics, no cherry-picked demos.
- 🔬 Core stack: deep learning · autonomous agents · quantitative pipelines.
- 🎧 Off the clock: same pattern-finding instinct, applied to music.
Also building with: HuggingFace · LangChain / LangGraph · Pandas / NumPy · LightGBM / XGBoost
| Project | Problem | Key Tech | Impact |
|---|---|---|---|
| Signal Desk – TSE Robo-Advisor | Naive ML on emerging-market data gets fooled by capital increases, trading halts, and non-stationary prices. | Python LightGBM FastAPI SciPy |
Deflated Sharpe Ratio evaluation catches selection bias — ships honest metrics, not cherry-picked curves. |
| Brain Tumor Classification | Classify tumor type from MRI scans under strict validation. | PyTorch Keras OpenCV |
~88% test accuracy, Colab-ready pipeline. |
| Project | Preview |
|---|---|
| Signal Desk Dashboard | 🎥 [add GIF/link to docs/assets/dashboard-preview.gif] |
| Next demo | 🎥 [link] |
Eyna Shabani · B.Sc. Computer Science, IAU Tehran West




