Mini Tech Challenge - Call Analyzer

This project implements the requested mini challenge:

• Accept a customer call transcript via a simple web UI (FastAPI).
• Use the Groq API to generate a 2-3 sentence summary and a sentiment (Positive / Neutral / Negative).
• Print and save results into call_analysis.csv with columns: Transcript | Summary | Sentiment.

📱 Screenshots

---

Files included:

• app.py - FastAPI app + simple front-end form.
• groq_client.py - wrapper that calls Groq and falls back to a simple rule-based method when Groq is unavailable.
• templates/ - Jinja2 HTML templates for the UI.
• requirements.txt - Python dependencies.
• sample_transcript.txt - a short sample transcript you can paste.
• .env.example - example showing how to set GROQ_API_KEY.
• call_analysis.csv - will be created when you run an analysis.

⚡ Quick start

1. Clone / project

cd mini_tech_challenge

2. **(Optional)** Create a virtual environment:
   ```bash
   python3 -m venv .venv
   source .venv/bin/activate

3. Install dependencies

   pip install -r requirements.txt

4. Run the app

   uvicorn app:app --reload

   Open http://127.0.0.1:8000 in your browser, paste a transcript, and click Analyze.

5. Where output is saved

   • After analyzing, results are appended to call_analysis.csv in the same folder.
   • The CSV columns are: Transcript, Summary, Sentiment.

6. Notes about Groq usage

   • This project uses the official Groq Python client and their Chat Completions API to request a JSON response (summary + sentiment). See the Groq Python repo for usage examples and installation instructions. citeturn1view0turn0search4
   • If the Groq call fails (no API key, network error, or model output can't be parsed), the app falls back to a deterministic short-summary + token-based sentiment heuristic so you can still demo the functionality.

7. Example curl test

   curl -X POST -F 'transcript=Hi, I was trying to book a slot yesterday but the payment failed and I received no email confirmation.' http://127.0.0.1:8000/analyze

References

• Groq Python client & examples. citeturn1view0
• Groq Chat / text API docs. citeturn0search17