Course help where students already ask
The bot brings assignment support into the CS 382 Slack workspace, letting students ask for guidance through a slash command or an app mention without switching to a separate tutoring interface.
- Built the Slack application with the asynchronous Slack Bolt SDK.
- Handled both explicit commands and conversational app mentions.
- Used thread history to keep follow-up answers connected to the discussion.
Grounding Gemini in course material
Helpful responses are constructed from the assignment manual, starter code, and test cases so the model can guide students toward the relevant material instead of answering from generic context alone.
- Attached assignment artifacts to Gemini prompts.
- Prompted the assistant to guide students without handing them completed code.
- Separated helpful and intentionally sarcastic response modes.
Slack-native community tools
Beyond AI assistance, the bot packages recurring workspace tasks into commands for anonymous communication and channel-level participation summaries.
- Implemented /help, /sarcasm, /anon, and /stat commands.
- Supported anonymous posts through a Slack modal with optional file uploads.
- Added concurrency limits, request rate limiting, and exponential-backoff retries around external API work.