Sohaib Siddique Butt

I'm a Master's student, specializing in Artificial Intelligence and Large Language Models (LLM). I enjoy transforming complex ideas into concrete tools that solve real-world problems.

AdeptDoc: RAG-based Document Q&A System - Built an end-to-end RAG pipeline to extract and chunk multi-column PDFs using PyMuPDF, generate semantic embeddings with Gemini, and implement FAISS-based vector search for fast, accurate document question-answering with LLM integration.

Le Chat Social Intelligence Platform - Developed and deployed a Model Context Protocol (MCP) using FASTMCP and Mistral-medium-latest to automate social media workflows, integrate LLMs with prompt engineering, and deliver a cloud-ready, end-to-end system forreal-time content intelligence.

Multimodal Agent - Built a real-time multi-agent system using Gemini 2.0 Flash, computer vision, and multimodal LLMs as a proof-of- concept for live image analysis and contextual compliance reasoning on NVIDIA L4 GPUs.
 

Biology Search Engine for NASA's bioscience publications - Used Gemini LLMs to scrape and summarize research articles, implemented semantic search with embeddings and cosine similarity, and built a knowledge graph to link publications, authors, and themes.
 

- Currently researching the integration of Large Language Models (LLMs) with Retrieval-Augmented Generation (RAG) for candidate profile analysis and exploring multi-agent collaboration with LLMs
- Applying and expanding knowledge in Machine Learning, Computer Vision, and Reinforcement Learning to design intelligent data retrieval and analysis methods.