LLMs for Historical Dataset Construction from Archival Image Scans

Backend for my 'Stateful AI over Low-Bandwidth Networks' patent (FR2511116). A hybrid RAG pipeline (SmolDocling, ChromaDB, Reranker) with SMS support, separating local VLM (perception) from the main LLM (generation).

Automated patent novelty analysis pipeline with SerpAPI search and LLM evaluation

End-to-end RAG pipeline: extract patent PDFs, index semantic chunks, answer questions with local LLMs + multimodal image analysis.

A Flask-based web application that enables interactive querying of patent documents using OpenAI’s LLM and the Lens.org API. It supports fetching patent metadata, claims, and other key information, and facilitates contextual Q&A with the patent using a custom prompt pipeline.

Automated document classification and data extraction pipeline using UiPath and Local LLMs (Ollama/Phi-3). Processes PDFs (Invoices, Patents, Project Applications) and converts unstructured data into structured CSV format.

A pipeline and Streamlit app for forecasting the growth of quantum subfields using research publications, patents, and funding data. Includes data collection, cleaning, labeling (LLM), and modeling (polynomial regression + ridge), with customizable weighted predictions and country-level insights. Developed to help quantum researchers