Agentic - AI Project
Gemini Autonomous Agent
In this project I used Gemini API, and added some tools like: read mail, send mail, perform web search, & get news.
Then I gave access of that tools / functions to Gemini Model, so that Gemini can use that all tools. And after that we can tell gemini to take actions on our behalf.
For example we can say: Hey, get my mails and give me update about all in a concise manner. Or Hey send email to user@example.com regarding this, and it will perform that action without we having to do anything.
Retrieval Augmented Generation (RAG) Projects
Custom Document Search & Chat
In this project I made a Flask app micro service in which we can upload our custom documents. And then they will be converted to Vector Embeddings. Then we use cosine similarity search to take nearest matching files, and then pass all that and user's query to LLM.
So then we upload documents then we can go to an endpoint and just ask any query related to our documents, and then relevant files and our query both will be given to LLM then we can have a conversation about our documents in detail.
Product Insight System
In this project we have products in our SQL database. Then we take products from there and convert them into Vector Embeddings, later used to retrieve relevant top products from 1000's of other products.
After we convert them into embeddings we can have a user query in natural language like: Give me price estimate of 1 Sony headphone, 2 Logitech Mouse & 1 Apple keyboard. This will then find products calculate total and give us answer.