About

An engineeringcompany, not anagency.

BYAK exists to work on systems where the consequences of failure are operational rather than cosmetic. That constraint shapes who we hire, how we scope work and what we are willing to promise.

01Position

Where we work.

Most software can afford to be wrong for an hour. Some cannot — because a warehouse stops, a payment reconciles incorrectly, or an engineer in the field gets the wrong instruction. Systems in that category have different requirements, and they are the ones we build.

In practice that means enterprise operations platforms, regulated customer applications, field and workforce systems, security-sensitive infrastructure and enterprise data platforms including Palantir Foundry. It also means we spend a disproportionate amount of the engagement on the parts nobody demonstrates: authorisation models, failure paths, migration rehearsals and observability.

We are a senior team by design. The people who scope your engagement are the people who build it, and there is no account layer between you and them. That constrains how much work we take on at once, which is a deliberate trade.

03What we hold to

Four positions thatshape everyengagement.

  1. 01

    Architecture is a business decision

    Where the boundaries sit determines what will be cheap and what will be expensive for the entire life of the system. That is a commercial fact expressed in technical form, and it deserves the attention a commercial decision gets.

  2. 02

    Security cannot be added later

    The findings that matter are structural: authorisation scattered across a codebase, trust boundaries that were never drawn, secrets in places nobody audits. None of them are fixable by a scan two weeks before launch.

  3. 03

    The handover is the product

    Software that only its authors can change is a liability with a nice interface. We deliver reasoning, documentation and skills alongside code, and we would rather be retained by choice than by dependency.

  4. 04

    Saying no is part of the service

    Some projects should not be built, some platforms should not be bought, and some rewrites should be refactors. Telling a client that costs us revenue occasionally and saves them far more.

04How we work

What clients notice first.

  • Senior engineers on the work, not on the pitch

    The people who scope your project are the people who build it. We do not staff engagements with a senior architect for the first two weeks and juniors afterwards.

  • Architecture before implementation

    Every engagement produces an explicit architecture with recorded decisions. It costs weeks up front and saves quarters later, and it is the single largest predictor of whether a system stays maintainable.

  • Security is not a separate phase

    Threat modelling, authorisation design and secure delivery practice belong in the architecture. Security added at the end is either expensive or cosmetic.

  • We work directly with technical decision-makers

    No account layer between you and the engineers. Technical decisions get made in the conversation where they arise, with the people who understand the consequences.

  • Delivery you can see

    Working software in a production-like environment every two weeks. Scope and cost stay visible, and stopping remains an option you can afford to take.

  • Built to be handed over

    Documentation, runbooks and pairing are part of delivery. The goal is that your team can own the system, and that you keep working with us by choice.

05Delivery process

Seven stages, every engagement.

The depth of each stage scales with the work. Whether a stage happens does not.

  1. 01

    Understand

    What happens
    Structured sessions with the people who own the business outcome and the people who run the current system. We map the process, the constraints and the failure modes you live with today.
    What you receive
    A written problem statement, a constraint register and an agreed definition of what success means in business terms.
    Why it matters
    Most failed projects solved a well-specified version of the wrong problem. This stage is where that gets caught, while it costs nothing.
  2. 02

    Analyse

    What happens
    Review of the existing estate: code, data, integrations, infrastructure, delivery practice and incident history. We test assumptions against evidence rather than accepting the current narrative.
    What you receive
    An assessment of the current state with the specific risks, dependencies and unknowns that will shape the design.
    Why it matters
    The gap between how a system is described and how it behaves is where projects lose their schedule. We find it before committing to a plan.
  3. 03

    Architect

    What happens
    Domain model, service boundaries, data design, security model and integration strategy — each significant choice recorded with its alternatives and consequences.
    What you receive
    Architecture documentation, decision records, a threat model and a phased delivery plan with a cost envelope.
    Why it matters
    Architecture determines what will be cheap and what will be expensive for the system’s whole life. It is the highest-leverage stage in the engagement.
  4. 04

    Build

    What happens
    Iterative delivery in vertical slices, each reaching a production-like environment. Code review, automated testing and continuous integration are non-negotiable rather than aspirational.
    What you receive
    Working, deployable software every two weeks, with a demonstration against real scenarios and a current view of scope and spend.
    Why it matters
    Working software is the only honest measure of progress. Regular increments keep scope decisions reversible while they are still cheap.
  5. 05

    Validate

    What happens
    Functional, performance, resilience and security validation against the requirements agreed in stage one — including the failure paths, not only the happy ones.
    What you receive
    Test results, performance characteristics under expected and peak load, and a security assessment with findings resolved or explicitly accepted.
    Why it matters
    Validation against the original business requirement is what separates "the code works" from "the system does its job".
  6. 06

    Deploy

    What happens
    Staged rollout with a tested rollback path, data migration executed against a rehearsal, monitoring and alerting live before traffic arrives.
    What you receive
    A production system, runbooks, operational dashboards and a support arrangement agreed in advance.
    Why it matters
    Deployment is a risk event. Rehearsed, reversible and observable turns it into a routine one.
  7. 07

    Improve

    What happens
    Post-launch measurement against the original objectives, incident review, technical debt tracking and a prioritised backlog of the next increments.
    What you receive
    A review against the business case, a maintained architecture record, and a plan for the next phase — or a clean handover.
    Why it matters
    Systems that are not deliberately maintained decay predictably. Planned improvement is far cheaper than the rebuild that replaces it.

06Company

The company, in facts.

Founded
2024
Based in
Vilnius, LithuaniaBYAK MB · company code 306897734
Working across
Europe and the United Kingdom
Contact
Contact formOne queue for every enquiry

BYAK holds no industry certifications at present. There is no row for them because we do not publish credentials we cannot evidence, and a badge is a poor substitute for work you can inspect.

Want to knowhow we wouldapproach it?

The fastest way to assess an engineering partner is to give them a real problem and see how they think about it. That conversation costs nothing.

An engineer reads every enquiryNo pursuit if we are not the right fit

Senior engineering for operations platforms, security architecture and Palantir Foundry delivery — for organisations where a wrong answer stops the business.

Based inVilnius, LithuaniaWorking acrossEurope · United Kingdom

Contact

Discuss your project

Project enquiries, security disclosures, applications and data requests all arrive through the form, in one queue. A person replies typically within one working day.