CodiLime makes the invisible work

Consider a software engineer waiting for a build to finish. The code is written. The idea is ready. But somewhere between the keyboard and the customer, a chain of tests, environments and dependencies has become a queue. Hiring another programmer might make that queue longer. The useful question is where the work gets stuck.

The expensive part is the waiting

For Nutanix’s Xi Epoch, a monitoring product now part of Prism, CodiLime worked on precisely this unglamorous territory. The assignment was a continuous integration system that engineers around the world could use and extend. Its published account describes moving workflows from Jenkins to CircleCI, collecting requirements and creating reusable build resources. It separated stages so work could run in parallel.

Then came a test framework developers could run locally, with mocked system components and Docker-based environments. Harjot Gill, then general manager of Nutanix Xi Epoch, credited the collaboration with cutting development cycle time by 50%. He also reported an increase of more than 80% in test coverage in crucial system areas. These are client-reported project results, with a specific scope. The interesting part is the mechanism: make feedback arrive sooner.

Before the networks, an auction that fizzled

CodiLime’s beginnings were considerably less tidy. Its founders’ early venture, ClicXS, tried reverse auctions. Users paid to reveal a product’s price; each click lowered the eventual purchase price. The programmers could build the mechanism. Persuading enough people to use it proved harder.

In a 2023 interview with Puls Biznesu, co-founder Marek Niedźwiedź recalled scant sales and money running short for development and marketing. He also described overambitious projects and elaborate tools with too little demand. The team’s technical ability had survived a business idea that did not. University of Warsaw connections helped bring early programming assignments, including work for NTT.

That distinction shaped the company founded in 2011. Customer work provided a business; internal product experiments eventually became separate companies, including Dealavo, deepsense.ai and Neptune. CodiLime concentrated on business-to-business services. There is a useful lesson here for technically gifted founders: a working invention still needs a buyer. Selling expertise can reveal which problems buyers already have.

A software company that speaks packet

Today its niche sits where software engineering meets networks, cloud infrastructure and security. Customers include networking vendors, semiconductor businesses, telecoms and software companies. Cisco, Juniper Networks, Equinix and Palo Alto Networks appear in its public client portfolio. The common requirement is engineering that understands the infrastructure underneath an application.

Mesh7 offers a concrete example. The startup wanted visibility into communication between distributed application workloads, and controls over sensitive data flows. CodiLime helped build its microservices security platform, contributing DDoS protection, observability integrations, policy-generation mechanisms and an Envoy-based approach. Its account also describes a Go control plane using a Kubernetes operator. VMware later acquired Mesh7. That acquisition is part of the customer’s history, not proof that a contractor caused it.

At Selector Software, the work addressed repeatability and security: automated build, test and deployment pipelines, Kubernetes-based cloud deployments, backup-and-restore procedures and infrastructure management across cloud providers. These jobs explain the positioning better than a list of programming languages. CodiLime sells the ability to connect application behavior with the network and operating environment that make it possible.

“helped us reduce our development cycle time by 50%”

Harjot Gill / Nutanix Xi Epoch
Client testimonial published by CodiLime

Buy a team, borrow a specialty

The commercial offering takes several forms: add specialists to an existing team, commission a tailored team, hand over an end-to-end project or bring in consultants temporarily. Services span product design, backend and frontend engineering, test automation, network infrastructure, data engineering and hardware offloading. This is a services business. Its customers’ subscription products should not be mistaken for CodiLime’s own SaaS catalogue.

For a buyer, the practical comparison is with hiring internally, using a broad software consultancy or engaging another networking specialist. The attraction is concentrated domain knowledge alongside delivery capacity. The trade-off is coordination: an outside team still needs access, clear decisions and a well-defined problem. A small, conventional website would leave much of this particular expertise unused.

Cost belongs in that decision early. Define the system boundary, agree on a measurable outcome and request a scoped estimate. A buyer can then compare proposals against the same deliverables, rather than competing descriptions of effort. For infrastructure work, engineering fees are only one input; cloud usage, hardware, integration effort and internal staff time can also matter. Nutanix’s cycle-time result suggests a sensible question to copy: which recurring delay would make a measurable difference if it shrank?

A useful brief would name the workloads, the supported network equipment, the deployment environments and the people responsible for approving changes. It would also specify a recovery plan. Automation can reproduce a mistake as faithfully as a correct configuration. The buying task is therefore partly about making assumptions visible before those assumptions spread across a system.

The next queue is forming around AI

The company’s public recruiting material supplies a smaller clue about how it expects to deliver. It advertises flexible hours, remote or office work, English lessons and support for professional certifications. A technical interview is followed by a discussion about fit for a particular project. For a services company, that matching matters: customers buy the capabilities of the people assigned to their problem, rather than the average of everyone on the payroll.

Waterland invested in August 2023, backing continued growth and a buy-and-build strategy in networking and telecommunications. In his Puls Biznesu interview, Niedźwiedź explained the appeal: financial backing and an international network, while CodiLime could remain the platform bringing other businesses together. Earlier investor discussions had not offered a direction the founders wanted.

The leadership has since changed. Grzegorz Rycaj succeeded Niedźwiedź as CEO in September 2025. In March 2026, CodiLime appointed Arindam Guha chief revenue officer, explicitly tying US expansion to demand for AI infrastructure, low-latency networking, observability and compute acceleration.

The engineering question remains refreshingly testable. At AutoCon 4 in 2025, CodiLime and eBay discussed lessons from network orchestration; a separate CodiLime workshop asked participants to diagnose failures in a live network of more than 200 nodes using AI. The exercise put a fashionable technology beside an old operational need: finding what broke. Whether the tool is a build pipeline or an AI assistant, the buyer can ask the same thing. Does it shorten the wait?