Proposal: Network Engineering SPE

Treasury
12

Period: 28 July 2026 – 28 August 2026

Status: On track

Source thread: Proposal: Network Engineering SPE

Summary: The pilot’s retrospective and full budget were published, two more retroactive grants were decided and merged, a subgraph audit was commissioned and is under way, and the next round of funding has been put to the community as a pre-proposal.

Completed Deliverables

  • Retroactive Grants — two decisions, two different builders. Both ran the published process end to end: application, AI rubric pre-screen, Review Team decision, published rationale.
    • Add a realtime Flux Klein example over Trickle (Moein / zargarzadehm) — approved, $500 in full. A self-contained FLUX.2-klein-4B realtime image-to-image example over Trickle and the Live Runner, with persistent session handling, live prompt and parameter updates, frame-skipping for low latency, health-gated model startup, and both offchain and paid onchain execution. Approved subject to the merge completing with the original contributor’s attribution preserved from PR #32. Application · Review Team decision · PR #35
    • vllm-realtime: realtime speech transcription over Trickle + WebSocket (Gideon) — approved, $500 in full. The network’s first transcription example: PCM audio in over Trickle, transcript streamed back word-by-word over an orchestrator-proxied WebSocket, with a GPU-free mock backend so the whole pipeline runs on a laptop and a real vLLM/Voxtral backend for inference. Independent review deployed it through a registered orchestrator; listed in the repository on 17 August. Application · Review Team decision · PR #73
  • Direct Grant #2 — Livepeer Subgraph Audit (BuildersDAO) — approved, audit under way. $6,500 USD-equivalent in LPT ($5,700 audit and technical review, $800 coordination and reporting), delivered as a single milestone: a prioritized audit report with findings tiered from critical production risk down to larger redesign. Commissioned after the Explorer subgraph incident that blocked delegator withdrawals and forced a multi-day resync. Audit and reporting only — remediation is scoped and priced separately once findings are ranked. Non-sensitive findings will be filed as public GitHub issues, with a closing forum retrospective. Applied 3 August, approved 6 August, community input session held 13 August. Application · Review Team decision
  • Direct Grant #1 — Livepeer Payment Clearinghouse (John Mull / Elite Code Solutions) — Milestone 3 signed off, Milestones 4 & 5 rescoped. PymtHouse went to public release on 4 August, with GitHub and Discord login, usage attribution aligned to the three go-livepeer 0.9.0 billing unit types, personal API keys, and a published builder integration guide. Milestone 3 closed on 20 August: Administrative API, Auth0 and OpenMeter setup scripts, SignerSession token exchange and user-scoped Usage endpoints. Milestones 4 and 5 were rescoped in the open to documentation and a retrospective, leaving the remote signer out of the livepeer/clearinghouse deployment rather than funding a second one. PymtHouse public release · Milestone 3 close-out · livepeer/clearinghouse
  • Pilot retrospective published — commitments scorecard and full budget. Eleven pieces of work funded across three mechanisms, from a $500 bounty to a five-figure Direct Grant, involving more than ten contributors and teams. Retrospective
  • Network Engineering Priorities roadmap session — 11 August. Walked the community through the retro, then opened the floor on which priority areas to fund and which parts of the mechanism to change. The output shaped the structure of the next round. Feedback session notes
  • Network Engineering SPE II pre-proposal published — 25 August. $230,000 USD-equivalent in LPT for 1 September to 31 December 2026. The mission is unchanged; the shape is not. Four tracks, each around a core ecosystem stakeholder, each with one named owner, one outcome by year end and a budget the owner controls: Easy to build (Mike Zupper), Easy to operate (Josh Allmann), Easy to delegate (Elliott Conway), Easy to validate (Shane Burgett). Open for feedback on the forum now. Pre-Proposal: Network Engineering SPE II
  • Review capacity — external reviewer engaged. A paid review bounty was bought from ECWireless on 4 August to clear the Explorer merge queue that had been waiting on a single reviewer’s calendar. The retro carries this as a learning, and SPE II budgets a dedicated reviewer line from day one. Budget