Hey there 👋

I'm Aayush Shah

AI Enthusiast | Problem Solver

A passionate problem-solver with expertise in machine learning, NLP, and full-stack development. I've built scalable systems, optimized workflows, and delivered high-accuracy solutions through coding, automation, and AI.

Let's connect and build something impactful together!

Let's connect! GITHUB

My last projects </>

RoleChat: AI Persona Creator
Beta Users, AI Character, Chatbot

This platform lets users create, customize, and chat with dynamic AI personas—from anime heroes to original creations—with seamless cross-platform continuity (web to Telegram). Creators can design personalities using a no-code studio and share creations in a community. Built for fans, writers, and roleplayers seeking immersive, adaptive storytelling experiences.

Agent Engine: AI-Powered Workflow Automation
Private

Developed a Python framework for building AI agent workflows that self-adapt to task requirements. Simplifies complex automation with dynamic task routing, multi-LLM orchestration, and stateful memory for context-aware execution. Features a modular architecture for adding custom agents/tools—ideal for creating research assistants, customer support systems, or data pipelines that evolve in real-time.

News LLM: AI News Curator
LLM,Scraping

Created an AI-powered news curator that automatically gathers, summarizes, and delivers real-time updates from global sources to platforms like Slack. By leveraging Large Language Models (LLMs), the tool distills complex articles into actionable insights, empowering teams to stay informed without manual effort. Designed for adaptability, it supports seamless integration of new data sources and communication channels.

Docbite: Information Extraction From Structured Documents

Engineered a smart document processing system to automate data extraction from tax and insurance forms using machine learning. Built an intuitive annotation tool to accelerate data preparation and a self-improving training module to boost accuracy over time. The platform simplifies high-stakes workflows, enabling businesses to process documents faster and with fewer errors.

Chatterbox: RAG Chatbot with Streamlit

Built an AI document assistant that lets users chat with PDFs, reports, or research papers. Powered by RAG, it provides context-aware answers and maintains conversation history for seamless follow-ups. Designed with Streamlit for simplicity, it streamlines tasks like customer support, research, or data tasks—all through a conversational interface.

Anubadak: Interpreter implemented in C++

Designed and built a lightweight programming language interpreter from scratch in C++, complete with modern features like function overloading, type flexibility, and scoped variables. Emphasized clean syntax and error handling to create a tool ideal for prototyping or educational use, demonstrating mastery of low-level language design principles.

RoleChat: AI Persona CreatorBeta Users, AI Character, Chatbot