Syed Muhammad Ali Fatmi

Worked on integrating FHIR with RAG to extract clinical data and used LLMs to generate accurate, context-aware responses for healthcare queries.

● Integrated large-scale FHIR medical datasets for semantic search and retrieval using Python, Pandas, and NumPy. ● Fine-tuned Llama 3.1 on GPU servers using CUDA and PyTorch Distributed (DDP) for high-performance experimentation. ● Applied Retrieval-Augmented Generation (RAG) pipeline to generate accurate, context-aware clinical responses. ● Developed FastAPI endpoints and containerized backends using Docker for reproducible experiments.

Worked with Large Language Models (LLama 3 7B, 13B, and 3.1 8B) for healthcare and e-commerce applications. Utilized Retrieval-Augmented Generation (RAG) with LLMs, fine-tuned OpenAI models, and applied prompt engineering for optimized outcomes. Contributed to IT audit projects leveraging AI solutions.

● Fine-tuned LLaMA 13B models on multi-GPU setups using PyTorch, CUDA, and NCCL for healthcare queries, improving answer accuracy by 90%. ● Deployed on the AWS cloud platform(EC2), ensuring scalability and real-time availability. ● Integrated with Lex and Polly for voice-based interaction, boosting patient engagement by 80%. ● Built an interactive chatbot enabling users to query e-commerce data in natural language. ● Developed RAG-based retrieval frameworks using LangChain and Vector Databases (Chroma, FAISS) for improved answer accuracy. ● Utilized prompt engineering to transform SQL query outputs into natural language explanations generated by LLMs. ● Developed an AI-powered IT audit assistant using LLM (LLaMA 3.1 8B) to automate key audit workflows. ● Applied RAG( retrieval-augmented generation) pipelines for faster scoping and achieved 70% quicker risk prioritization. ● Automated audit report generation, reducing preparation time by 24 hours per report.

started as a machine learning intern which will expand my practical skills also will be familiar with professional environment