Rohit Kumar Yadav

📊 AI-Powered Retail Sales Forecasting & Decision Support

An end-to-end AI platform built with an FDE mindset to solve real-world retail business problems through sales forecasting, GenAI-powered analytics, and automated ML workflows.

• Built a Databricks Bronze–Silver–Gold data pipeline and a 30-day sales forecasting model with MLflow experiment tracking and model versioning.
• Developed a GenAI analytics assistant for natural-language business queries and integrated LangSmith for LLM tracing and observability.
• Implemented FastAPI backend APIs, a React + Vite dashboard, and workflows for data uploads, pipeline execution, and model retraining.
• Containerized the application services with Docker and deployed them on AWS EC2, using NGINX to load-balance traffic across multiple backend instances and GitHub Actions to automate CI/CD workflows for streamlined builds and deployments.

✈️ Aircraft Maintenance Intelligence System

AI-powered aircraft maintenance platform that analyzes flight telemetry, detects anomalies, retrieves maintenance manual evidence, and generates grounded maintenance recommendations.

• Analyzed 14 aircraft parameters using historical statistics, Z-score anomaly detection, trend analysis, and risk classification.
• Built maintenance-manual retrieval covering normal ranges, warning/critical thresholds, failure modes, symptoms, causes, troubleshooting, inspections, and maintenance actions.
• Implemented LangGraph orchestration + Groq LLM for evidence-grounded maintenance reasoning.
• Integrated LangSmith for end-to-end LLM workflow tracing, debugging, and observability.
• Developed an interactive React + Vite dashboard with risk insights, anomaly alerts, parameter analytics, maintenance recommendations, and manual evidence.
• Productionized with FastAPI, Docker, GitHub CI/CD, Vercel, and Render.

✈️ Agentic Trip Planning System

AI-powered travel planning platform that creates personalized multi-day itineraries using multi-agent workflows, real-time APIs, and LLM-driven recommendations.

• Built a multi-agent workflow using LangGraph and Groq LLMs.
• Integrated weather, hotel, attractions, geolocation and transportation APIs.
• Generated personalized day-wise itineraries based on user preferences.
• Implemented trip history, Markdown export and LangSmith monitoring.
• Deployed production-ready application on Render.

🤖 Agentic AI Assistant with MCP

Intelligent AI assistant powered by LangGraph, MCP, Groq and Ollama with autonomous tool execution.

• Integrated 7+ tools for AI-powered automation.
• Built custom MCP servers using SQLite and remote integrations.
• Supported expense tracking, weather, stocks, web search and local events.
• Implemented tool orchestration and autonomous decision making.
• Used LangSmith observability for debugging and monitoring.

🛡️ RAGGuard — LLM Evaluation, Reliability & Observability Framework

A component-wise and lifecycle-oriented evaluation framework for RAG and LLM applications. RAGGuard evaluates retrieval quality, generation quality, end-to-end RAG performance, application quality, security & safety, operational reliability, regression behavior, and production performance.

📚 Multimodal AI System

Advanced RAG platform for querying PDFs, YouTube videos and multiple document sources.

• Indexed 1000+ document chunks for efficient retrieval.
• Implemented Corrective RAG architecture.
• Built intelligent routing using LangGraph.
• Improved retrieval accuracy and response quality.
• Deployed publicly using Streamlit Cloud.

✍️ Agentic Blogger

Multi-agent AI platform for autonomous blog generation, research and content creation.

• Automated over 80% of the content creation workflow.
• Integrated web search and citation generation.
• Implemented persistent memory using PostgreSQL.
• Reduced content creation time by 75%.

🚗 Car Price Prediction API — Production-Ready ML Backend

An end-to-end Machine Learning application that predicts car prices through a secure, scalable REST API. Built with production-focused features including authentication, caching, monitoring, automated testing, CI/CD, Docker, and AWS EC2 deployment.

• Built a high-performance REST API using FastAPI with Pydantic request validation.
• Implemented JWT authentication, custom middleware, request logging, and performance tracking.
• Integrated Redis caching to optimize repeated prediction requests.
• Configured Prometheus and Grafana for API monitoring and observability.
• Developed automated tests using Pytest to validate application behavior.
• Containerized the application with Docker and automated CI/CD using GitHub Actions.
• Deployed the ML backend on AWS EC2 for cloud-based inference.
• Implemented the ML inference pipeline for preprocessing, price prediction, and evaluation.

☀️ Your Morning AI Buddy

Real-time AI assistant delivering weather, news summaries and personalized recommendations.

• Supports 50+ cities with real-time weather updates.
• Implemented LLM-powered news summarization.
• Delivered personalized recommendations.
• Deployed publicly using Streamlit Cloud.