Raspberry Pi Agents — LinkedIn Job Discovery + Coffee Scout
Two agents running on a Raspberry Pi: one surfaces fresh LinkedIn job postings, another monitors specialty coffee shops for new beans.
The Setup
Hardware: Raspberry Pi (sub-$35) Model: Kimi 2.5 (instead of Claude) Cost: $0.50 per extensive test run (vs $5 burnthrough on Sonnet)
Agent 1: LinkedIn Job Discovery
LinkedIn's algorithm deprioritizes recent posts in favor of promoted listings — this agent cuts through:
- Finds buried jobs: Surfaces postings on page 4+ that the algorithm hides
- Fresh listings: Prioritizes recent posts over promoted ones
- Fully autonomous: Runs 24/7 on the Pi
Agent 2: Coffee Scout
Custom agent monitors specialty coffee shops for:
- New bean arrivals
- Limited releases
- Roaster updates
Why Kimi 2.5?
| Model | Cost per test run |
|---|---|
| Claude Sonnet | $5 |
| Kimi 2.5 | $0.50 |
10x cost savings with comparable performance for these tasks.
The Philosophy
Raspberry Pi = fully autonomous, sub-$35 hardware, zero cloud dependency.
This is the "edge AI" model: small, cheap, local, always-on agents that don't need cloud connectivity to function.
Source
- Original post by @sanketpath (6 likes)
- February 20, 2026