I'm an AI/ML Engineer passionate about building intelligent systems at the intersection of AI and science. I'm currently building Notes9 — an AI-native Electronic Lab Notebook (ELN) that helps researchers and scientists work smarter.
- Currently Building: Notes9 — AI-native ELN with embedded AI assistant (Catalyst), RAG pipelines, and molecular viewers
- Focus Areas: LLMs, RAG systems, AI Agents, Agentic Workflows, Medical AI
- Background: Fracture Detection (YOLO/DETR), Medical Prognosis, Clinical NLP, Biomedical NLI
- Currently Learning: Multi-agent orchestration, AWS Bedrock, Mastra agent framework
- Ask me about: AI/ML, RAG pipelines, Claude Code agents, Next.js + Supabase, Python
- Fun fact: I built an agent that routes other AI agents by task — and learns from your ratings
Languages
Frontend & UI
Backend & APIs
Database & Cloud
AI / ML
AI Techniques
| Project | Description | Stack |
|---|---|---|
| Notes9 | AI-native ELN — Catalyst AI assistant, RAG pipelines, molecular viewers, Cytoscape research map | Next.js 16, Supabase, FastAPI, AWS Bedrock, TipTap |
| agent-router | Routes Claude Code agents by task using BM25 + bandit algorithm — learns from your ratings | JavaScript, Claude Code |
| Fracture-Detection | Multi-class bone fracture detection with YOLOv4, Faster R-CNN & Detection Transformers | Python, PyTorch, DETR |
| LLMs-in-Health-Science | MedBERT, MedRoBERTa & Longformer ensemble for clinical trial NLI (SemEval 2024 Task 2) | Python, HuggingFace |
| Attention-Mechanism | Full Transformer from scratch — encoder, decoder, multi-head attention + positional encoding | Python, PyTorch |
| Home Credit Risk | Credit default prediction with LightGBM, SHAP explainability & MLflow experiment tracking | Python, LightGBM, SHAP |
| Chess + RL | DQN vs MCTS in PettingZoo chess env — 4672 action discrete space | Python, PyTorch, PettingZoo |
| CTR-Extraction | LLM-powered SAP clinical trial analyzer with FDA E9 compliance checking | Python, LLMs, NLP |
| Language-Translation | Fine-tuned seq2seq models EN↔GR and RO↔EN with Streamlit UI | Python, Transformers, Streamlit |
