Featured Projects

Technologies: Python | React | FastAPI | ChromaDB | LangChain

I engineered a RAG pipeline and architected a containerized deployment for a specialized LLM. This involved implementing Parameter-Efficient Fine-Tuning (PEFT) using LoRA to adapt the model to specific domains, ensuring high-performance inference through a FastAPI backend and a React/Chainlit frontend.

I also specified an evaluation framework using DeepEval and 'LLM-as-a-judge' methodologies to systematically measure the model's faithfulness and reasoning capabilities.

Technologies: ROAR Collab HPC | Slurm | LangChain

As part of my university research, I am modeling a RAG-based UI that leverages the LionPath API to provide citation-grounded model responses. This requires utilizing HPC resources and job schedulers for heavy computations.

The system is targeted to achieve over 80% retrieval accuracy, utilizing rule-based safeguards and escalation protocols to triage academic inquiries reliably.

Technologies: ASP.NET Core | PostgreSQL | React | Material UI

I developed an ASP.NET Core Web API utilizing a service-oriented architecture for scalable data management. The platform integrates the Google Gemini API and NewsAPI to aggregate and summarize content, delivered through a dynamic React frontend.

To ensure security, I established robust user authorization and role-based access control using JWT tokens across the RESTful endpoints.