README.md

InterviewAlly

Note: This is an experimental app.

InterviewAlly is an AI-powered interview assistant. It uses Retrieval-Augmented Generation (RAG) to ingest an interview handbook and answer questions contextually using Groq LLMs and local embeddings.

Features

• RAG Architecture: Retrieves context from a local markdown knowledge base.
• LLM Provider: Uses Groq for high-speed inference.
• Embeddings: Uses local HuggingFace Transformers (all-MiniLM-L6-v2) to avoid external embedding costs.
• Interface: Includes a simple web UI and a REST API.

Technologies

• Runtime: Node.js, TypeScript
• Web Framework: Express
• AI Framework: LangChain
• LLM: Groq
• Vector Store: In-Memory
• Embeddings: HuggingFace Transformers

Prerequisites

• Node.js (v18 or higher recommended)
• A Groq API Key

Installation

1. Install dependencies:

   npm install

2. Prepare the Knowledge Base: Ensure you have your source material located at:

   data/interview-handbook.md
   

3. Environment Configuration: Create a .env file in the root directory with the following variables:

   LLM_API_KEY=your_groq_api_key
   LLM_MODEL=llama3-8b-8192

Running the Application

Start the server:

npx ts-node src/server.ts

The server will start at http://localhost:3000.