AI engineer in the making · Data Science undergraduate at SLIIT, Sri Lanka
I build LLM agents, retrieval systems and computer vision models, end to end.
- 🎓 BSc (Hons) in Information Technology, specialising in Data Science, at SLIIT
- 💼 Former AI/ML intern at Mobitel, Sri Lanka
- 🤖 Working on multi-agent LLM pipelines, hybrid-retrieval RAG and computer vision
- 🏠 Prefer local models through Ollama, so the systems I build stay cheap to run
- 🔎 Open to junior AI / ML engineering roles, remote or with relocation
- 📫 Reach me through the contact form on my portfolio or LinkedIn
| Project | What it does | Built with |
|---|---|---|
| X-Ray Bone Fracture Detection | Two-stage pipeline: a classifier screens each X-ray, then YOLOv8 localises the fracture only when one is flagged. AUC 0.90, recall 0.74 on FracAtlas, with Grad-CAM explanations. | PyTorch, EfficientNet-B0, YOLOv8, Grad-CAM |
| JARVIS Voice Assistant | A LangGraph voice agent with 50+ tools for PC and phone control, an intent router, FAISS long-term memory and face-authenticated admin mode. | LangGraph, Ollama, FAISS, MCP, FastAPI |
| AskGit | Ask plain-English questions about any GitHub repository. Hybrid retrieval fuses dense embeddings and BM25 with Reciprocal Rank Fusion. | LangChain, ChromaDB, BM25, FastAPI, React |
| DefenseOSINT (team) | Multi-agent OSINT pipeline. I led the security audit against the OWASP Top 10 for LLMs and STRIDE and found 8 vulnerabilities. | Multi-agent, Ollama, FastAPI, React |
➡️ More projects and details on my portfolio.
AI & retrieval: LangGraph · LangChain · RAG · MCP · Ollama · FAISS · ChromaDB · Pinecone · LaBSE · Ultralytics YOLO
- IBM: Introduction to Deep Learning & Neural Networks with Keras
- IBM: Machine Learning with Python
- IBM: Introduction to Software Engineering
- EDUCBA: Mastering Kafka Consumer Systems for Telecom Data
- Udemy: Java Spring Boot Microservices · Java Programming Masterclass
