Indie dev replaces $1k OpenAI bill with a YAML pipeline on spot GPUs and vLLM
cmm324 · reddit · 2026-10-21
A Reddit user cut his nightly product-catalog AI pipeline bill ($1k on OpenAI over months) by moving to spot GPUs in Seoul with vLLM and an open model, then spent months building a proper scheduler: pipelines written as one YAML doc, per-step schema validation, tool calls to your own endpoints, human-in-the-loop steps, and the whole graph run on spot GPUs within a deadline up to a week. Example support-desk flow: 10 nodes, $0.0089 per item. He's recruiting people with nightly/weekly batch AI jobs to test it for $20 in credits.
More from coding & agent
- OpenAI Live API tip: control voice sessions from your backend via sideband WebSocket — pbbakkum · 2026-10-21
- OpenAI's MultiLangBench: PHP scores highest (91.3%) at building a database engine from scratch — ricklamers · 2026-10-21
- UserProof lets Claude Code watch your user session recordings and boost conversion for $29/mo — doooyle · 2026-10-21
- dphnAI's distributed inference network served 32B DeepSeek tokens in 24h, ~6% of Venice's daily volume — Ronangmi · 2026-10-21
- Superconductor revamps agent picker spanning 8 coding agents with search and favorites — sergeykarayev · 2026-10-21
- Metaphor claims cloud coding agent beats Claude Code with 25% fewer tokens — Scobleizer · 2026-10-21