End-to-end offline OCR and semantic parsing pipeline for identity documents based on YOLOv11, PaddleOCR, and LLaMA-3.1.

ReHGNN is a deep learning framework for intelligent server recommendation in distributed large-model deployment scenarios.

Production RAG pipeline — grounded retrieval, source-cited answers, Precision@k + MRR eval. CLI + Flask REST API. Gemini · ChromaDB · Python 3.11+

A robust, modular, and production-ready platform for solar power system data analysis, machine learning, and prediction. Built with ZenML, Streamlit, MLflow, and a rich Python data science stack, this project enables end-to-end workflows from data ingestion and EDA to model training, deployment, inference and experiment tracking

🦾 MATA | Model-agnostic computer vision framework — detect, track, segment, classify, embed & more. One API for HuggingFace, ONNX, and TorchScript models. Built-in ByteTrack/BotSort tracking, ReID, VLM agent tool-calling, barcode decoding, and OCR evaluation. Zero-shot ready. 🚀

Self-hosted model switcher for agents with profile-based provider accounts, encrypted credentials, OpenAI-compatible proxying, and per-agent manual model selection.

The MultiGen pathfinding model aims to predict a suitable model invocation path based on user input requirements, using known model information.

Semantic runtime above Citadel for raw language intake, context assembly, model orchestration, semantic journaling, reply publication, and restart-safe intent synthesis into structured policy decisions across the stack.

A production-ready, frugal, sovereign AI system that orchestrates India's open-source language models to achieve state-of-the-art reasoning on consumer hardware through Test-Time Compute (TTC) and Cognitive Serialization.

Build a reliable Retrieval-Augmented Generation pipeline that ingests documents, stores vectors, and generates grounded, source-cited answers.