README.md

metadata

title: ai_vidya Pipeline Agent
emoji: 🤖
colorFrom: purple
colorTo: blue
sdk: docker
app_file: app.py
pinned: false

ai_vidya_bot

Slack agent for the AI Department intern pipeline. Manages Week 0 through Probation (W1 + W2) in #ai_vidya.

File structure

ai_vidya_bot/
├── main.py          # Flask app + Slack event endpoint
├── router.py        # Event classifier and dispatcher
├── handlers.py      # Four event handlers
├── scheduler.py     # APScheduler cron jobs
├── sheets.py        # Google Sheets read/write layer
├── parser.py        # Report format parser and quality scorer
├── prompts.py       # Claude evaluation prompts (Week 0, W1, W2)
├── requirements.txt
└── .env.example

Setup

1. Google Sheets

Create a Google Sheet with these 7 tabs (exact names):

• candidates
• reports_log
• signals_log
• warnings_log
• commitments
• evaluations
• decisions_log

Add headers to each tab matching the column names in sheets.py.

Create a Google Service Account, download the JSON key, and share the sheet with the service account email.

2. Slack app

Create a Slack app at https://api.slack.com/apps

Required OAuth scopes:

• chat:write
• channels:history
• im:write
• im:read
• users:read

Enable Event Subscriptions. Set the request URL to: https://your-space.hf.space/slack/events

Subscribe to bot events:

• message.channels

Install the app to your workspace and invite it to #ai_vidya.

3. HuggingFace Spaces

Create a new Space (SDK: Docker or Gradio).

Add all variables from .env.example as Secrets in the Space settings.

Upload the service_account.json file and set GOOGLE_SERVICE_ACCOUNT_PATH to its path in the Space filesystem.

4. Deploy

pip install -r requirements.txt
gunicorn main:app --bind 0.0.0.0:7860

Adding a candidate

POST to /admin/add_candidate:

{
  "slack_user_id": "U12345",
  "name": "Arjun Mehta",
  "cohort_id": "COHORT_001",
  "ft_slack_id": "U67890",
  "hr_slack_id": "U11111",
  "week0_start_date": "2025-05-12"
}

Logging a Week 2 commitment

POST to /admin/log_commitment after the Day 6 FT conversation:

{
  "candidate_id": "CAND_001",
  "description": "Streaming component working and tested",
  "due_date": "2025-05-21",
  "due_time": "EOD"
}

Report format (candidates post this in #ai_vidya)

Name: [full name]
Date & Time: [now]
What I worked on: [description]
What I got stuck on: [specific blocker or 'nothing today']
What I tried to resolve it: [what you attempted]
Current status: [where things stand]

Cron schedule (IST)

Time	Job
9:00 AM	Morning FT brief
12:30 PM	Missed report check (noon window)
5:30 PM	Missed report check (5pm window)
6:00 PM	EOD signal questions to FT
6:30 PM (Wed)	Deadline check — Wednesday deliverable
12:00 PM (Fri)	Deadline check — Friday deliverable
8:00 PM	End-of-stage evaluation (Day 5 + Day 10)
10:30 PM	Missed report check (10pm window)