Cloud cost optimisation
and FinOps skills

FinOps is the practice of managing cloud spending as a shared responsibility. The FinOps Foundation describes it as an operational framework and cultural practice that maximises the business value of cloud through collaboration between engineering, finance and business teams.¹ Core skills include cost visibility and allocation (tagging, accounts, showback and chargeback), forecasting and budgets, removing waste (idle and orphaned resources), right-sizing compute, storage and databases, scheduling non-production environments, using commitments and spot capacity wisely, storage lifecycle policies, unit economics (cost per order or user), managing AI and GPU spend and building engineering accountability. Cloud's pay-as-you-go model, as AWS explains, makes costs flexible but easy to lose track of.²

The field is changing fast. The State of FinOps 2026 report found 98% of practitioners now manage AI spend, up from 31% two years earlier, and 78% of FinOps teams report to the CTO or CIO.³ It also notes diminishing returns from early optimisations as teams move from big wins to many smaller ones. In India, AmbitionBox lists FinOps analysts at a typical ₹5.5–6.1 lakh from a small sample,⁴ and cloud architects, who often own cost-efficient design, at ₹28.6–31.7 lakh.⁵

FinOps teams managing AI spend (State of FinOps 2026)

98%

Share two years earlier (State of FinOps 2026)

31%

FinOps teams reporting to CTO or CIO (State of FinOps 2026)

78%

Typical FinOps manager pay in India (AmbitionBox, small sample)

₹16.7–18.5 lakh

Table: How cloud cost optimisation skills are assessed (editorial summary)

Table: Sequencing cloud cost optimisation (illustrative)

Table: How FinOps developed (editorial summary)

• A tagging policy and allocation report
• A waste analysis with savings estimates
• A right-sizing recommendation with data
• A commitment analysis
• A unit cost metric
• A verified cloud cost optimisation certificate

Show savings and accountability: "Introduced tagging and team showback; removed idle resources and right-sized databases; cost per order fell by a third while orders grew."

"How do you start optimising a cloud bill?"

Visibility first — tags and allocation — then waste, right-sizing and finally commitments.

"When should you buy commitments?"

After removing waste, for predictable baseline usage, with coverage and risk reviewed regularly.

"How do you get engineers to care about cost?"

Show them their own costs and unit costs, make savings easy and celebrate wins.

"What's showback vs chargeback?"

Showback reports costs to teams; chargeback bills them internally.

"How would you manage AI spend?"

Track GPU and AI service usage by team and feature, set budgets, optimise model choices and monitor unit costs per request.

• "Is spend allocated by team?" — visibility.
• "Who owns cost decisions?" — accountability.
• "How large is AI spend?" — priorities.
• "Are unit costs tracked?" — maturity.

Table: Cost-related roles and typical pay in India (AmbitionBox, October 2026; indicative)

Table: Common cloud cost tools (editorial guidance)

• Cloud engineer / financial analyst — foundations.
• FinOps analyst — reporting and recommendations.
• FinOps practitioner / engineer — optimisation programmes.
• FinOps manager — organisation-wide practice.
• Head of cloud / CTO office — strategy.

1. Week 1: Visibility — Tag and allocate costs in a sample account.
2. Week 2: Optimise — Find waste and right-size; estimate savings.
3. Week 3: Prove it — Define unit costs; take the Cloud Cost Optimization Certification.

Related: Cloud Computing, Unit Economics and Site Reliability Engineering. Guides: cloud computing, unit economics and SRE.

• Reports — weekly team costs.
• Anomalies — investigating spikes.
• Recommendations — right-sizing and cleanup.
• Commitments — coverage reviews.
• Forecasts — with finance.
• Coaching — engineers on cost-aware design.

"What's spot or preemptible capacity?"

Spare capacity at a discount that can be reclaimed — good for fault-tolerant batch work.

"How do you detect cost anomalies?"

Budgets, anomaly detection tools and daily trend checks by team.

Table: A first-90-days plan (editorial guidance)

Start with visibility: who spends what and why. Deliver quick wins on waste to build credibility. Then introduce unit costs and a commitment strategy so savings last.

AI workloads have become the fastest-growing area of cloud spend, and the State of FinOps 2026 report shows nearly all practitioners now manage it. GPU instances, model inference and AI services need their own tracking, budgets and unit costs — such as cost per AI request or per document processed. FinOps teams also use AI tools to detect anomalies and recommend optimisations. Practitioners who understand both AI architectures and cost models are in high demand.

Export billing data and group costs by service, account, team tag and environment. Look at month-over-month changes to find the biggest increases. For each, ask what changed: new workloads, traffic growth or waste? Check utilisation of the largest compute and database resources. Look for unattached storage, old snapshots and idle load balancers. Compare non-production and production spend. Summarise findings with estimated savings and owners. A clear bill analysis is a strong portfolio piece.

"Commitments are the first step." They lock in waste if you buy them too early.

"Cost is finance's job." Engineers make most cost decisions through design.

"Cheaper is always better." The goal is value — cost per outcome, not lowest spend.

Allocation turns one large bill into numbers each team can act on. Start with account or project structure — separate accounts per team or environment make allocation simple. Then enforce a small, mandatory set of tags such as owner, team, environment and product, using policies that block untagged resources or reports that flag them. Shared costs, such as networking or security tools, need a fair split, often proportional to each team's usage. Publish the method so teams trust the numbers. Showback reports, sent monthly or weekly, are usually the first step; chargeback can follow once teams are confident in the data.

• No tagging.
• Buying commitments before cleanup.
• Ignoring non-production spend.
• Totals without unit costs.
• Savings without owners.

FinOps trends come from the State of FinOps 2026; definitions from the FinOps Foundation; cloud pricing concepts from AWS; and India pay from AmbitionBox (read October 2026; small samples for FinOps roles). The optimisation sequence, bill analysis guidance and history are editorial. Science of Apps publishes the free test linked here. Last reviewed October 2026.

• FinOps — collaborative cloud financial management.
• Right-sizing — matching resources to actual need.
• Commitment discount — lower prices for committed usage.
• Showback — reporting costs to teams.
• Unit cost — cost per business outcome.