Built a scalable ecommerce platform using microservices architecture with gRPC for inter-service communication and GraphQL as the API gateway. Includes services for account management, product catalog, and order processing. Each service has its own database with Account and Order services using PostgreSQL and Catalog service using Elasticsearch for optimal performance.

A dynamic AI chatbot that handles multi-step tasks and contextual conversations using LangGraph agents and Groq inference. It combines LangChain orchestration with Streamlit UI for a clean and powerful workflow.

A voice agent that can answer questions and provide information using natural language processing.

Developed and deployed a full-stack AI agent capable of querying databases using natural language. Built using Next.js 16 with API Routes for serverless backend, React frontend, and Vercel AI SDK integrated with OpenAI GPT models for tool-calling and SQL generation. Utilized Drizzle ORM with Turso (SQLite) and implemented a real-time chat interface using SSE for streaming AI responses.

AI-driven NLP system to analyze a TV series using LLMs for extracting story themes, sentiment, and character insights. Scraped dialogues using Scrapy, applied Transformer models to detect emotional tone and thematic clusters, and visualized character relationships using spaCy NER, NetworkX, and PyViz. Developed a character chatbot with fine-tuned Hugging Face models for realistic dialogue simulation.

Built an LLM-powered chatbot over 50+ news articles using a Retrieval-Augmented Generation (RAG) pipeline with Jina embeddings and Qdrant vector DB. Implemented session-based memory with Redis and REST/WebSocket APIs for real-time conversational flow. Designed an interactive React + SCSS UI with streaming bot replies and optimized chat performance.

Learned and implemented end-to-end workflow from web scraping to LLM fine-tuning using Python, Decodo, and Hugging Face. Built reliable web crawlers with Decodo's API supporting JavaScript rendering, proxy rotation, and anti-blocking. Cleaned and structured scraped datasets, then fine-tuned models like LLaMA for news classification tasks, covering text classification, generation, and evaluation.

A real-time AI assistant that sees, hears, thinks, and responds using webcam, audio, and advanced LLM tools like LangGraph and Groq. Dora is an agentic experience built for immersive interaction.

Step-by-step tutorial showing how to design, build, and orchestrate multi-agent AI workflows with CrewAI. Demonstrates agent roles (researcher, writer, fact-checker, automator), message routing, tool calling, and evaluation loops to mimic real-world teams for research, writing, and automation tasks.

A resume-aware chatbot using RAG, LLM function-calling, and LangChain that provides contextual responses.

AI-based tool that generates LinkedIn posts in a user's writing style using few-shot learning.