Overall Position:

You will build and run GreenNode's unit-economics model as its hands-on owner: the cost model, the definitions, the allocation rules and the monthly reconciliation to the general ledger. The CFO is accountable for the number; you build it, run it monthly and can defend every line.

Responsibilities:

1. Cost and unit economics (~75%) Cost model and definitions. Map the GL into cost pools across all product lines — hardware depreciation by SKU/server/site; data-centre capacity commitments (space, power incl. PUE, cooling); network, storage, interconnect and egress; platform and shared services (IAM, tenancy, billing, metering, observability); support, security and compliance; resilience capacity and expected service credits. Build and maintain the asset register (SKU, server, rack, site, in-service date, useful life, capex). Prepare useful-life and COGS-vs-opex policy options for the CFO, who decides.

Unit costs, published monthly. Cost per available and per occupied GPU-hour by cluster and SKU; cost per vCPU-hour and per instance shape on vServer ; cost per TB-month by storage tier; cost per Gbps and per TB egress; cost per million input and output tokens, and blended cost per million production tokens by model and service tier. Contribution by customer, workload, product line and configuration.

Three cost views, kept separate.

Marginal — incremental cost of serving more usage within existing capacity, for tactical acceptance and floor pricing. Fully-loaded at actual utilisation — this month's economic truth including unused capacity, for contribution and board reporting. Fully-loaded at target utilisation — the rate card and planning view, with unabsorbed capacity as its own line, consistent with IAS 2 "normal capacity".

Allocation of shared platform and support costs. This is a judgment call with real consequences: the basis you choose decides which business looks profitable. You will propose it, defend it, document it and keep it stable.

Capacity earning its keep. Track utilisation and distinguish capacity that is simply unsold from capacity that is stranded because no instance shape or workload fits what is left.

Reconciliation and cadence. Pools → drivers → products → reconcile to the ERP trial balance , every month, to the dong. Variance commentary; quarterly refresh of the standard/target rate card; annual useful-life review.

Decision support. Cost floors for pricing and discount authority; SLA and resilience cost priced into commitments; capacity and procurement cases (buy vs lease, cost per unit at actual vs target utilisation).

Later, not day one. Design how allocations are posted back into the ERP at standard rates with a separate unabsorbed-capacity account, with the CFO and Group Accounting — only after the definitions survive two clean quarterly refreshes.

1. Planning, reporting and business partnering (~25%) Monthly management reporting — revenue, cost, margin and the operating KPIs behind them — and the variance commentary. Support the annual plan and monthly forecasts. Partner with Product, Engineering, Commercial and Operations on pricing discussions, investment cases and profitability by product line.

5–8 years in FP&A, infrastructure or capacity finance, management accounting or transaction services, with hands-on ownership of a cost allocation or TCO model in an asset-heavy business (cloud, data centre, telecom, logistics, utilities, or manufacturing with real activity-based costing). Genuine command of cost allocation and capacity economics: you can explain without hesitation why unit cost rises when utilisation falls, and how idle capacity should be treated. Fluent in depreciation, capex and utilisation economics; able to write a definition that Engineering, Product and Group Accounting will all sign. Reconciles to the ledger. A model that does not tie to the trial balance is not finished. SQL and advanced Excel; comfortable pulling from a data warehouse and from operational systems (metering, capacity, billing). Applied AI in daily work — Claude, Copilot, ChatGPT or equivalent — to accelerate analysis, reconciliation and commentary. Works through partnership: you get your inputs from engineers and sales ops, not from a ticket queue. Strong communication with both technical and non-technical audiences. English working proficiency; Vietnamese a strong plus.

Nice to have Cloud, GPU or data-centre experience: GPU-hours, kWh × PUE, rack-kW, instance shapes, tokens per GPU-hour, cache-hit rates. Costing a shared or multi-tenant service, where one asset serves many customers and attribution is not obvious. FinOps Foundation unit-economics practice, FOCUS cost/usage schema, TBM-style cost pools. CMA / ACCA / CPA / CIMA; Oracle ERP and HFM; IFRS / VAS.

This is not A manufacturing cost accountant role (standard costing, BOM, inventory variances) — welcome, but not sufficient. A customer-facing "cloud economics" / TCO pre-sales role. A cloud-bill FinOps analyst role (tagging, reserved-instance coverage) — our COGS is our own infrastructure, not a hyperscaler invoice.