https://github.com/gokularaman-c/ai-interview-assistant
An AI-powered interview assistant that performs real-time transcript summarization, pause-aware questioning, and adaptive interview evaluation using transcript and screen content.
https://github.com/gokularaman-c/ai-interview-assistant
ai fastapi frontend full-stack human-computer-interaction interview-assistant machine-learning natural-language-processing real-time-systems speech-processing
Last synced: 26 days ago
JSON representation
An AI-powered interview assistant that performs real-time transcript summarization, pause-aware questioning, and adaptive interview evaluation using transcript and screen content.
- Host: GitHub
- URL: https://github.com/gokularaman-c/ai-interview-assistant
- Owner: gokularaman-c
- Created: 2026-01-10T09:02:11.000Z (5 months ago)
- Default Branch: main
- Last Pushed: 2026-01-10T09:58:37.000Z (5 months ago)
- Last Synced: 2026-01-11T03:12:13.932Z (5 months ago)
- Topics: ai, fastapi, frontend, full-stack, human-computer-interaction, interview-assistant, machine-learning, natural-language-processing, real-time-systems, speech-processing
- Language: HTML
- Homepage:
- Size: 9.77 KB
- Stars: 0
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# NavGurukul AI / Full Stack Hackathon — Challenge 1 + Challenge 2
This repository contains two working MVPs built for the NavGurukul hackathon.
---
## Challenge 2: Ultra-Lightweight Client-Side Interview Assistant (Offline)
* Runs fully in the browser (HTML + JavaScript)
* Simulates STT via a transcript textbox
* Live extractive summary updates
* Pause detection + contextual filler question generation
* Filler spoken using browser TTS (SpeechSynthesis API)
* Performance metrics shown (page load, summary update, filler generation, pause detection)
* Works offline with no backend dependency
---
## Challenge 1: AI-Driven Automated Interviewer (Backend)
* FastAPI backend with endpoint: `POST /interview`
* Accepts:
* `transcript` (simulated speech-to-text)
* `ocr_text` (simulated screen / slide / code OCR text)
* Performs:
* Context analysis from transcript + screen content
* Dynamic interview question generation
* Follow-up question generation
* Scoring across multiple dimensions
* Actionable feedback generation
---
## Integrated Demo
The frontend (Challenge 2) is integrated with the backend (Challenge 1).
Clicking **Generate Interview** sends transcript + OCR text to the FastAPI backend and displays:
* Next interview question
* Follow-up question
* Score
* Feedback
---
## ▶️ How to Run
### 1️⃣ Backend (Challenge 1) — Port 8001
```bash
cd server
pip install -r requirements.txt
uvicorn app:app --reload --port 8001
```
Swagger UI:
* [http://127.0.0.1:8001/docs](http://127.0.0.1:8001/docs)
---
### 2️⃣ Frontend (Challenge 2) — Port 8000
```bash
cd client
python3 -m http.server 8000
```
Open UI:
* [http://localhost:8000](http://localhost:8000)
---
## Demo Steps
1. Paste transcript lines into **Transcript** box (simulating live STT stream)
2. Paste screen / slide / code content into **OCR Text** box
3. Click **Start**
* Live summary updates
* Pause detection + filler questions via TTS
4. Click **Generate Interview**
* Backend generates interview questions, score, and feedback
---
## API Reference
### POST `/interview`
**Request**
```json
{
"transcript": "string",
"ocr_text": "string"
}
```
**Response**
```json
{
"next_question": "string",
"follow_up_question": "string",
"score": {
"technical_depth": 0,
"clarity": 0,
"originality": 0,
"implementation_understanding": 0
},
"feedback": ["string"]
}
```
---
## ✅ Status
* Challenge 1: **Completed**
* Challenge 2: **Completed**
* End-to-end demo: **Working**
---
## Author
Gokularaman C