DeveloperDocs AI Copilot
End-to-end RAG system to analyze and answer questions on developer documentation. Uses vector embeddings, semantic search, and LLM generation for accurate, context-aware answers.
12 projects spanning LLMs, RAG systems, MLOps, NLP, and full-stack AI applications
End-to-end RAG system to analyze and answer questions on developer documentation. Uses vector embeddings, semantic search, and LLM generation for accurate, context-aware answers.
Experimenting with the CrewAI Agent Framework for multi-agent orchestration and autonomous task execution. Explores agent roles, task delegation, and collaborative AI workflows.
Building intelligent agents using the OpenAI Agents SDK. Explores agent creation, tool integration, and autonomous task execution with OpenAI's latest agent framework.
Exploring LangChain to create custom LLMs, deploy models, and integrate APIs. Covers prompt templates, chains, memory, agents, and tool usage within the LangChain ecosystem.
ATS checker application for matching resumes against job descriptions. Uses Google's open-source LLM to analyze resume-JD fit, with a Streamlit frontend for interactive results.
End-to-end blog generation using two approaches: AWS Bedrock LLM and Llama 2 from Hugging Face. Generates blogs based on topics and word count with configurable model outputs.
Multi-capability chatbot using Gemini LLM with Q&A, image analysis, text embedding, and conversational capabilities. Integrates Google's open-source models via API.
Image analysis application using Gemini Vision Pro LLM. Users upload images and ask questions, with the LLM providing detailed analysis and seamless interaction via Streamlit.
End-to-end ML system with data ingestion, transformation, and model training pipelines. Implements CI/CD with GitHub Actions, Docker containerization, AWS ECS deployment, and model monitoring.
MS capstone project: Fine-tuned FLAN-T5 (10B parameters) on the Stanford Alpaca dataset. Compared full fine-tuning vs PEFT with LoRA, evaluating with ROUGE metrics against human baselines.
Implementing sentence transformers with embeddings and multi-task learning. Explores SBERT models for semantic similarity, text classification, and embedding-based applications.
React application with a searchable user list fetching data from an API. Demonstrates component architecture, TypeScript typing, state management, and real-time search filtering.