Run any open-source LLMs, such as DeepSeek and Llama, as OpenAI compatible API endpoint in the cloud.

The collaborative spreadsheet for AI. Chain cells into powerful pipelines, experiment with prompts and models, and evaluate LLM responses in real-time. Work together seamlessly to build and iterate on AI applications.

A holistic way of understanding how Llama and its components run in practice, with code and detailed documentation.

An Automated AI-Powered Prompt Optimization Framework

Drop in replacement for the OpenAI Assistants API

[ICCVW 25] LLaVA-MORE: A Comparative Study of LLMs and Visual Backbones for Enhanced Visual Instruction Tuning

AI agent with RAG+ReAct on Indian Constitution & BNS

PasLLM - LLM inference engine in Object Pascal (synced from my private work repository)

A lightweight Python API wrapper and CLI for Perplexity’s Sonar language models.

Multimodal RAG based on Llama 3.2

This repository contains projects developed to showcase how to apply Generative AI and open-source LLMs in the construction industry

AI Screen Analyzer allows users to capture screenshots, analyze them using various AI providers and models, and engage in conversations about the images.

The GroqCloud API wrapper for Delphi provides access to models from Meta, OpenAI, MistralAI and Google on Groq’s LPUs, offering chat, text generation, image analysis, audio transcription, JSON output, tool integration, and content moderation capabilities.

Simple RAG system powered by Milvus.

Generate an optimal path on a map a la Traveling Salesman Problem to hit different landmarks in an input city using Cloudflare Workers AI (using LlaMA-3.1), LangChain (for prompt templates and comma separated list output parser), Mapbox to create a map and get information about cities and landmarks in those cities, and Folium to edit the map.

Cloud platform for building AI agents

An example of integrating local LLMS using mlc-llm into an iOS app

ResurrectAI is an AI-driven chat application designed to bring the wisdom and knowledge of great historical personalities to life. Leveraging advanced language models and fine-tuning techniques, ResurrectAI enables users to interact with AI avatars of iconic figures, gaining access to their insights, guidance, and philosophical teaching in realtime

AI Engineering: Annotated NBs to dive into Self-Attention, In-Context Learning, RAG, Knowledge-Graphs, Fine-Tuning, Model Optimization, and many more.

Import Clickup Docs and chat with a local llama

A Python base cli tool for tagging images with joy-caption-pre-alpha models.

Voz-AI is an AI-powered RAG based voice assistant that is designed to help customers inquire about queries and retrieves the most accurate and relevant information back to them. This project was made for the AIQoD Hackathon 2025, eventually winning us the runner-up position.

A minimal agentic app to interact with OLLAMA models leveraging multiple MCP server tools using BeeAI framework.

SOC Analyst Automation using a RAG model integrates a knowledge retrieval system with generative AI to automate SOC Level-1 tasks. It processes server logs, retrieves relevant security insights, and generates accurate responses, enhancing incident analysis, reducing response times, and improving efficiency in handling cybersecurity threats through

"Explore the cutting-edge Conversational AI project, leveraging advanced NLP & ML to create intelligent chatbots. Dive into code for natural language understanding, dialogue management, and dynamic response generation, designed for scalability and seamless integration across platforms."

A sophisticated red-teaming agent built with LangGraph and Ollama to probe OpenAI's GPT-OSS-20B model for vulnerabilities and harmful behaviors. (Specifically built for the OpenAI Open Model Hackathon)

Generate personalized cold emails based on job postings and send them to potential clients or recruiters.

📧AI assistant for email automation, data retrieval, and managing company records with precision and efficiency.

Web Client For Ollama - Llama LLM

codebase and pipeline to generate high-quality synthetic datasets using Llama 3.1 and an innovative constitutional ruleset framework.