HackerRank had a problem that a cheaper server would not solve. As its developer-assessment business grew, its build-release pipeline introduced delays and downtime. AWS expenses were climbing. The cloud was doing what clouds do: making it easy to buy more capacity while leaving the awkward questions about how to use it to somebody else.
- Cloud-cost software, automation, and engineers in one offering.
- Optimizes both what you pay and what your workloads consume.
- Now connects shared infrastructure and AI costs to customers.
DoiT’s response began inside the pipeline. It recommended faster Docker container startup, parallel unit tests, and moving the build server to Amazon ECS with AWS Fargate. For non-production Auto Scaling groups, it introduced Spot optimization with a fallback to on-demand capacity. The published case reports a 24% reduction in EC2 expenses and nearly 50% less on-demand EC2 spending in those non-production groups. Those percentages describe different scopes. Neither means the entire AWS bill was halved.
Non-production Auto Scaling groups
That distinction explains a good deal about DoiT. A discount changes the price of a resource. An engineering change can alter how much of that resource a business needs, how quickly it ships, and how reliably it serves people. DoiT sells a combination of those interventions.
02The dashboard comes with people
Founded in 2011 by Vadim Solovey and Yoav Toussia-Cohen, DoiT grew out of cloud expertise and procurement. Today, its Cloud Intelligence platform combines multicloud cost analytics, allocation, anomaly detection, forecasting, and automation with technical guidance. Its own site reports more than 4,500 customers. The audience includes FinOps practitioners, platform engineers, and finance teams trying to make the same bill answer different questions.
For an engineer, the question might be which workload changed. For finance, it might be which product should carry the expense. For a SaaS executive, it might be whether a large customer is profitable. Cloud Analytics brings billing data together; Allocations assigns it to useful business groupings; DataHub adds outside business and usage data. CloudFlow turns repeatable responses into workflows. The point is to make cost information usable inside ordinary work.
DoiT’s Forward Deployed Engineers add architecture reviews, troubleshooting, migration guidance, and help implementing changes. Its public pricing distinguishes free Essentials from paid Enhanced and Enterprise plans. Enhanced includes expert inquiries; Enterprise adds a named engineer, procurement advice, and enterprise service commitments.
03Two kinds of waste, one expanding toolbox
The company has been buying its way deeper into the machinery. PerfectScale, acquired in 2025, brought Kubernetes optimization: adjusting container resources while paying attention to performance and resilience. LiveDiagrams added infrastructure visualization. CloudWize extended the portfolio into security posture and compliance. In January 2026, DoiT announced the acquisition of SELECT, bringing Snowflake cost and performance optimization into its orbit.
These purchases address a practical split. Rate optimization asks whether you can pay less for the capacity you buy, through commitments and purchasing strategy. Usage optimization asks whether the workload needs that capacity at all. A well-priced, oversized cluster still wastes money. An efficient cluster can still be bought badly. DoiT’s portfolio spans both questions, with data-platform optimization extending the work beyond virtual machines.
Its market position sits between cloud-native billing tools, independent FinOps platforms, automation specialists, and consultancies. AWS Cost Explorer or a provider’s own dashboards may be enough for straightforward reporting. IBM Apptio Cloudability and CloudHealth are alternatives for cloud financial management; CAST AI is an alternative in Kubernetes optimization. DoiT’s proposition is the combination of software, execution help, and procurement expertise. Buyers should compare the actual workload and service scope, rather than count dashboard features.
04The customer who costs more than they pay
In July 2026, DoiT introduced Attribute, aimed at a problem that AI makes especially uncomfortable. Several customers can share a cluster, a database, or a gateway to a model provider. The infrastructure bill records consumption. It does not automatically explain which customer or feature caused it. Attaching a label to a server cannot settle every argument about who used what.
Attribute uses an eBPF sensor to observe runtime activity and connect consumption to customers, features, and agents. DoiT says it requires no SDK or application code changes. The promise is a more detailed cost-to-serve picture, including AI tokens and shared infrastructure. That can inform pricing, account renewals, and decisions about expensive features. It is a product claim to validate against your architecture, rather than a reason to assume every environment will behave identically.
“You can’t tag a customer in a multi-tenant environment.”
Omri Cohen / Director of Engineering, Platform / Island
Island’s platform engineering director Omri Cohen puts the underlying difficulty plainly: “You can’t tag a customer in a multi-tenant environment.” Revenue arrives with a customer’s name attached. Infrastructure expense often arrives without one. Connecting the two can change which growth looks attractive.
05A price tag, and a test worth copying
There are two prices to understand here. One is the price of DoiT. Its website lists bespoke pricing for paid Cloud Intelligence tiers. An AWS Marketplace listing offers a 12-month Enhanced contract at $42,000, with additional usage terms. Treat that as a published purchase option, not a universal quote. The second price is the internal engineering time needed to investigate, approve, and implement changes. Include both when calculating whether the savings justify the purchase.
DoiT’s own capital story requires similar care. Charlesbank and affiliated investors put more than $100 million into the company in 2019, its first outside investment. The $250 million announced in March 2025 was a fund for acquisitions. It should not be added to the earlier investment as though it were another disclosed fundraising round.
The HackerRank example offers a useful sequence to copy: identify the operational bottleneck, change the workload, then optimize the purchasing mechanism. Start with a bounded environment and measure cost alongside release speed and reliability. Spot capacity suits workloads designed to tolerate interruption; it needs different handling when interruption would break a customer promise. Commitment discounts deserve caution when demand is unstable. A small, predictable bill may not justify another platform and service contract.
DoiT’s remote working model fits that emphasis on responsibility: its careers material describes asynchronous decisions, documented work, and judging output rather than time online. The customer version of that idea is concrete. Someone must own the change after the chart appears. A bill can tell you that money left the building. The useful work begins when someone can explain why, decide whether it was worth it, and safely do something about it.