showcase.
Engineered systems.
featured.
pip install mutant-ai
|
View on PyPI
|
Contribute & Star on GitHub ★
Most AI teams struggle with having the right evaluation data. Mutant AI solves this by taking your existing dataset and mutating it into an n-dimensional space of test cases, allowing you to evaluate your system across every possible dimension. For adversarial testing, you can also deploy its Red Team Agents to autonomously hunt for vulnerabilities, hallucinations, and edge cases using hypothesis-driven attack vectors.
1,500+ Downloads — currently building the next version, adding an Evaluator (similar to DeepEval & Ragas) to evaluate RAG systems and Agents.
Tech: Multi AI Agent, LLM-as-a-Judge, Adversarial AI, Python
college & experimental.
Systems and tools built during my college years, for learning and weekend explorations.
A lightweight, iterative research agent that breaks down complex questions, searches the web, extracts structured evidence, and synthesizes a cited report.
Tech: Ollama, Local LLM, Tavily, Python
A strategic decision-support system designed to protect seller margins and enhance customer trust through data-driven logistics intelligence.
Tech: MLflow, Docker, Machine Learning, Python
Automates writing SOAP notes after consultations by transcribing conversations using Deepgram and generating structured notes with Gemini.
Tech: LLMs, FastAPI, Deepgram, Gemini, Python
A cloud-native, GitOps-driven re-architecture modernized with Docker, Kubernetes, CI/CD, and AWS EKS, It is my old project API-Nexus, i just Applied DevOps on it for practice.
Tech: Kubernetes, AWS EKS, Helm, GitHub Actions, ArgoCD
Developed a multi-agent system for intelligent extraction and validation of insurance claims from complex Loss Run Reports, Build this for Pibit.ai
Tech: FastAPI, CrewAI, Gemini Flash 2, Python
Offers accessible legal assistance focused on Indian law through a multi-agent AI system powered by CrewAI and RAG.
Tech: Streamlit, CrewAI, Gemini Flash 2, FAISS
An AI-powered system analyzing NFL games, enabling real-time insights and decision-making using Multi-Agent collaboration.
Tech: CrewAI, Ollama, Llama 3.2, Python
An open-source, local-first full-stack project for experimenting with the evaluation and post-training of medical language models.
Tech: Local LLM, Multi-Agent, Evaluation, SFT / DPO, Python
A conversational AI agent built for answering geographical queries and providing location-specific insights.
Tech: LlamaIndex, FAISS, Python, Llama 3.1
An AI-powered tool for generating concise news summaries from lengthy articles.
Tech: Phidata, Python, Llama 3.2
A Retrieval-Augmented Generation (RAG) system built to extract intelligence and insights from the 2024 Union Budget Speech.
Tech: Langchain, Cassandra, Llama, Python
pet projects.
Before GPT 3.5: Built entirely by hand, without the assistance of modern generative AI.