Ankita Mishra's Work | ContraWork by Ankita MishraAnkita Mishra
AI Engineer: RAG ⢠AI Agents ⢠Document AI ⢠LLM Application
This is basically a tutorial for beginners who want to learn LangChain, I have curated relevant sources and provided my understanding in here to help learners move fast, and dont need to dig hours for relevant study material.
I hope this helps! JOYI is an AI assistant built using LangGraph, Streamlit, and Google Gemini. It supports memory, streaming responses, and a clean ChatGPT-style interface. Built a role-aware conversational AI assistant for manufacturing operations, enabling users to query and analyze structured database data using natural language.
Implemented Text-to-SQL, LangGraph agent orchestration, RBAC, conversational Human-in-the-Loop clarification, controlled database write operations, confirmation workflows, audit logging, and dynamic tables/visualizations.
The system intelligently asks for clarification when requests are ambiguous and safely executes authorized actions only after validation and user confirmation.
š This project is an agentic lesson-generation system that creates a beginner lesson, evaluates it against hard pass/fail quality checkpoints, and regenerates it when the lesson fails. The workflow terminates when the lesson passes or the retry budget is exhausted.
The lesson is designed for a 12th-grade graduate from India with limited English vocabulary, a non-English-medium background, and no prior AI knowledge.
š