Project TBD Memory Vault

NVIDIA Engagement Structure — Draft v0

4 min read

NVIDIA Engagement Structure — Draft v0

Companion to problem-framing bullets: that doc is the cover; this is the operational shape of the engagement. Both go to NVIDIA together.

Operating principle

Solve the underlying problem, not the surface symptom. The dashboard NVIDIA scoped from the EY workshop is the visible ask; the latent problem — how NVIDIA understands, communicates with, and services its customers and how they use NVIDIA’s products — is what we are actually here to address. Every workstream below ladders to that.

Legal — close by Wednesday next week

  • Mutual NDA between Bliss + Dustin and NVIDIA covering NVIDIA’s reverse supply chain structure, operational processes, software systems, and customer support data.
  • NVIDIA-facilitated (or umbrella’d) NDAs with customers, contract manufacturers, and 3PL logistics partners. These are structural, not optional: the data required to solve the problem lives across these parties, and an independent operator is the only credible collector.
  • All NVIDIA-sourced data remains perpetually owned by NVIDIA; standard data-retention terms.

Discovery — 14–30 day sprint

NVIDIA-side interviews (target 6–10):

  • Customer Service Operations lead — case lifecycle and intake (Lonny’s org)
  • Field Service / On-site engineer — actual repair dynamics in customer DCs
  • Demand Planning specialist — replenishment vs. return tracking
  • IT Solutions Architect, customer service ops — SFDC ↔ SAP ↔ data lake plumbing
  • Quality / Failure Analysis engineer — failure modes, telemetry, prediction
  • Greg and Lonny remain executive sponsors throughout

Operational data needed for honest scoping:

  • Customer breakdown by units deployed and by revenue — share and prioritization
  • Failure/RMA breakdown by customer and SKU — does workload mix change wear patterns?
  • Current SLA attainment baseline on standard RMAs (today: unmeasured)
  • 90-day escalation log

Customer interviews (target 4–6, anonymized where required):

  • Hyperscaler: Hardware Operations Manager, Site Reliability lead
  • Government / national lab: procurement + facilities lead
  • Neocloud: operations lead

CM repair partner interviews (target 3–4):

  • Repair Operations / Aftermarket Services lead — runs the repair line
  • NVIDIA Account Manager at the CM — runs the relationship

3PL logistics partner interviews (target 3–4):

  • Reverse Logistics Operations Manager
  • Warehouse Manager at the NVIDIA-dedicated facility
  • NVIDIA Account Director at the 3PL

Product development — 14–45 day sprint, overlapping discovery from day 14

  • Forward-deployed-engineer model: Bliss embedded with Lonny’s customer-service ops team in Santa Clara, weekly product-feedback loop.
  • Dustin running discovery, partner outreach, and the shared problem-understanding doc in parallel.
  • Deliverable by day 45: working prototype — designs, engineering, QA, internal deployment.
  • Working partnership with:
    • NVIDIA Solutions Architects for cross-system integration
    • NVIDIA IT / Infra for staging and deployment access
    • Greg and Lonny for product-direction sign-off

Problem understanding — continuous, the through-line

  • Live shared document, co-owned by us and Greg, codifying the underlying problem(s) as discovery sharpens them. This is the artifact that protects against bandaid-stacking.
  • Biweekly review with Greg + Lonny.
  • Quarterly playback to broader NVIDIA leadership — gives the engagement executive air cover and surfaces additional sponsors.

Financial — milestones with satisfaction triggers

  • Phase 0 (signing, NDA, scoping): completed at contract execution.
  • Phase 1 (Discovery): milestone payment on delivery of the shared problem-understanding doc + interview readout, target day 30.
  • Phase 2 (Prototype): milestone payment on Greg/Lonny sign-off of working prototype, target day 60.
  • Phase 3 (Operating relationship): standalone Bliss + Dustin operating company runs the data layer on an ongoing commercial basis; terms agreed in writing by Phase 2 completion.
  • IP: Bliss + Dustin own the underlying platform and code; NVIDIA owns NVIDIA-sourced data perpetually.
  • Satisfaction trigger at each phase: NVIDIA can exit cleanly if not satisfied. We carry the risk of phase-to-phase performance, which is what allows us to argue for the long-term operating relationship on the back end.

Out of scope (v0)

  • Telemetry collection from in-production NVIDIA units at customer sites — acknowledged as structurally blocked. Future-state work, not Phase 1–2.
  • Government-market reverse logistics — separate workstream once commercial pattern is proven.
  • Front-end customer web-portal redesign — we plug into existing customer portal surface, do not rebuild it.

Anchor docs: latent problem codification; problem-framing bullets; source meetings Greg + Lonny @ NVIDIA, 2026-06-17 and Joe Malchow advice, 2026-06-17.

Graph View

Backlinks