class AmineManai:
def __init__(self):
self.role = "Actuarial and Data Science Engineer"
self.next_step = "M.Sc. Actuarial Science @ Le Mans University 🇫🇷 (Sept 2026)"
self.focus = ["Machine Learning", "Deep Learning",
"Recommendation Systems", "Multi-Agent AI", "RAG"]
self.stack = ["Python", "PyTorch", "Scikit-learn", "LangGraph", "Next.js"]
self.current = "Building end-to-end ML systems: data → model → API → demo"
self.open_to = "Summer 2026 internship (remote preferred 🌍)"
def motto(self):
return "Turn data into decisions — and research into products."- 🎓 Major of my class (17.02/20) — Data Science & Actuarial Science @ ESPRIT
- 🔬 I love reproducing research papers and pushing them beyond the original results
- 🏅 NVIDIA Deep Learning Institute certified · Cisco CCNA
- 🗣️ Arabic (native) · French (C1) · English (C1)
- 📫 Reach me: amine.manai@esprit.tn
| Project | Description | Tech | Highlight |
|---|---|---|---|
| 💹 Trady | Forex education platform with multi-agent AI + multimodal RAG tutor | Django LangGraph Ollama FAISS |
4 specialized agents, local RAG with cited answers |
| 🖼️ CaptionAI | Image-captioning research reproduction + web app | PyTorch CLIP Next.js |
BLEU-4 = 26.51 on COCO (beats the original paper) |
| ⚽ WC26 Predictor | World Cup 2026 match prediction & Monte Carlo simulator | XGBoost Scikit-learn Next.js |
CRISP-DM on 30,710 matches · log-loss 0.8675 |
| 🌳 RLT | Reinforcement Learning Trees — hybrid ML algorithm | Python Flask MLOps |
94.15% avg accuracy on 10 UCI datasets |
| 🚕 Uber Analytics | Pricing prediction, segmentation & geospatial maps | Scikit-learn Folium Flask |
R² = 77.2% · K-Means demand zones |
| 🏭 SomipemERP | Business-process digitalization + KPI dashboards | Next.js NestJS SQL |
Manual process → structured digital product |
🔗 More on my portfolio →
