AI/ML Tools: AWS Bedrock, AWS Lex, Amazon Connect, Langchain, LangGraph, NLTK, H2O GPT, Rasa
Large Language Models (LLMs): Claude, Llama 2/3/3.1
Embedding Models: Cohere, AWS Titan
Cloud Technologies: AWS (EC2, S3, Lambda, Secrets Manager, RDS, IAM, EC2 with GPU), Google Colab 1
Other: Linux, Docker, Container Registry
Databases: SQL, NoSQL, VectorDB (Pinecone, ChromaDB)
Version Control: Git
Video Transcription (Q&A Bot):
Description: Developed a Question Answering system using Retrieval Augmented Generation (RAG), combining LLM foundation models with vector databases to enhance QA Bot accuracy and efficiency. The setup was entirely serverless, featuring a data cleaning and ingestion pipeline that splits text chunks and generates metadata.
Domain: Natural Language Processing, Question Answering Bot
Technologies Used: AWS Bedrock, AWS Lambda, S3, Secrets Manager, Pinecone, Python, RAG, Titan embeddings
LangChain Agentic Tool Calling Bot:
Description: Built an intelligent agent using Langchain to automate tasks by seamlessly integrating with external tools and APIs, streamlining workflows and boosting overall productivity.
Domain: Conversational AI, Task Automation
Technologies Used: AWS Bedrock, OpenAI, Python, Langchain, Cohere Embeddings, Knowledge base, Pinecone/ChromaDB
LangGraph Multi-Agent Customer Support System:
Description: Designed and implemented a multi-agent customer support system using LangGraph to handle customer queries efficiently and provide personalized responses.
Domain: Conversational AI, Customer Support
Technologies Used: AWS Bedrock, Python, LangGraph, Cohere Embeddings, Knowledge base, Pinecone/ChromaDB
RASA Chatbot:
Description: Developed a chatbot using the Rasa framework to automate customer interactions, provide instant support, and enhance customer satisfaction.
Domain: Conversational AI, Customer Service
Technologies Used: AWS Bedrock, Python, LangGraph, Cohere Embeddings, Knowledge base, Pinecone/ChromaDB
Omihub (Multi-Input Multi-Domain AI Agent):
Description: Built a versatile AI agent capable of processing input from multiple sources (text, voice, etc.) across different domains, showcasing proficiency in complex AI system development.
Domain: Artificial Intelligence, Natural Language Understanding
Technologies Used: Python, FastAPI, ReactJS, Django, NLTK, Cohere Embeddings, Knowledge base, Pinecone/ ChromaDB/ RDS/ DynamoDB/ MongoDB/ Sqlite, PostgreSQL
Freight Planner Bot (SQL Query + RAG + API):
Description: Developed a freight planning bot that leverages SQL queries, RAG, and API integrations to provide optimized freight itineraries and recommendations for air and sea transport.
Domain: Logistics, Freight Planning
Technologies Used: Python, SQL, RAG, React, FastAPI, LangGraph, Cohere Embeddings, Knowledge base, Pinecone/ChromaDB/RDS
NLTK-Based Question Answering System:
Description: Created a bot using NLTK to process user input prompts, search for matching questions in a dataset, and deliver accurate and relevant answers.
Domain: Natural Language Processing, Question Answering
Technologies Used: Python, NLTK
H2OGPT Exploration and Fine-tuning:
Description: Conducted exploration and fine-tuning of H2O GPT for various text generation tasks, gaining hands-on experience with state-of-the-art language models.
Domain: Generative AI, Language Modeling
Technologies Used: Python, H2O GPT, Llama 2/3/3.1