Enabling easy LLM and Marketplace access to Farmers and those unfamiliar with Modern Tech

This project empowers farmers and tech-unfamiliar individuals by providing easy access to LLMs and an online marketplace via phone calls. Users simply talk to an AI in their local language to list products, get orders, and fulfill them. Authentication is simplified, using only a phone number, with the call itself serving as verification. This brings the power of AI and e-commerce to everyone, no complex tech needed.

Milestone Plan

• Week 1 Ideation and Setup

  • Discussion: Minutes of Discussion
  • Ideas:
    • ✅ SMS LLM
    • ❌ Marketplace MCP
  • Initial Plan: 6 week plan
  • Research Results: Indian Languages Translation Model Text To Speech Model

    (Aditya)

  • Setup: Call Interface Setup

    (Prajwal)

• Week 2 Core Infrastructure

  • Integrate API gateway with call interface, Check Demo

    (Prajwal)

  • AWS Setup
    • Pre-trained LLM API(Gemini 2.5 Flash) and language translation

      (Aditya)

    • Speech language detection for language translation model

      (Aditya)

    • Basic speech-to-text and text to speech

      (Shubham)

    • Integration of all models

      (Shubham)

    • Reproducibility of AWS Setup using AWS CloudFormation

      (Shubham)

  • In Depth testing of the Lambda Function

    (Akash)

• Week 3 Function Calling & Caching

  • Set up CI pipeline on github focusing on marketplace security

    (Akash)

  • Simple caching system & FAQ database for common farming questions

    (Aditya)

  • Mock Marketplace Frontend

    (Aditya)

  • Mock Marketplace Backend

    (Shubham)

  • Create an OpenAI client with function calling for the mock marketplace

    (Prajwal)

• Week 4 Integration

  • Implement OpenAI Client in old lambda function and integrate with Marketplace function calling

    (Prajwal)

  • Detect incoming audio language in lambda function

    (Shubham)

  • Add manual caching method

    (Akash)

  • Integrate caching system with OpenAI Client

    (Aditya)

  • Create unit tests & CI pipeline

    (Prajwal)

• Week 5 Improve & Deploy

  • Deploy Marketplace, Caching on AWS and resolve any errors

    (Shubham)

  • Use cases presentation

    (Akash)

  • Send phone number to lambda function endpoint

    (Prajwal)

• Week 6 Buffer Week

  • Modify github workflow to use public ip address of the deployed AWS EC2 instance

    (Prajwal)

  • Create CD pipeline for instant deployment on AWS

    (Prajwal)

  • Implement lambda function in EC2 instance and resolve any errors

    (Shubham)

  • Cache queries in lambda function corresponding to phone number for context, delete when call ends

    (Shubham)

  • Send sms to phone number on order received and create unit test for it

    (Aditya)

  • Modify cache get method for semantically similar queries

    (Prajwal)