Skip to content

Est. 2026Onshore engineering for operations

Security you can
reproduce.

We do the security work that has to hold up under examination: penetration testing and adversarial validation, operational technology and ICS security, compliance engineering, digital forensics and incident response, and cyber ranges. Findings arrive with a reproduction path attached, not as a scanner export with the severities left in.

Start
Fixed-fee discovery, 2–4 weeks
Estimate
A range, with its assumptions
Ownership
Your code, your cloud, day one
Exit
30 days' notice, any reason
  • TypeScript
  • React
  • Next.js
  • Python
  • Go
  • .NET
  • PostgreSQL
  • Kafka
  • Terraform
  • Kubernetes
  • AWS
  • Azure
  • Snowflake
  • dbt
  • React Native
  • FHIR
  • EDI X12
  • OPC UA
  • OpenTelemetry
  • Modbus
  • IEC 62443
  • Zeek
  • Ghidra
  • NIST 800-171

What we’re for

A finding you cannot reproduce is an opinion. A control you cannot evidence is a hope.

This division finds out whether a system is actually secure, proves it to whoever is asking, and answers what happened when it was not. Security is engineered into everything the software division builds — testing whether that succeeded is a separate practice on purpose, so the people checking the work are not the people who did it.

  • We’ll tell you not to build it

    Roughly one in five inquiries ends with us recommending an off-the-shelf product, an internal hire, or doing nothing for another year. Being wrong about this once costs us a project. Being wrong about it consistently costs us the business.

  • You own everything

    Code, infrastructure, accounts and documentation, in your name from the first commit. No escrow, no runtime licence, no clause that makes leaving expensive.

  • Senior people, small teams

    Four experienced engineers beat twelve inexperienced ones on work like this, and cost less. The people who sell the engagement are the people who deliver it; there is no bait-and-switch to a junior bench, because we don’t have one.

Industries

We learn one domain properly before we claim it.

Six sectors we know well enough to be useful in the first meeting, including which of your constraints are regulation and which are convention.

How it works

Five phases. You can stop after any of them.

Every phase has a defined exit, and the exit is always yours to take. That constraint is what keeps the work honest.

  1. Phase 1

    First call

    45 minutes · no charge

    One conversation to work out whether this is a problem we should be working on together.

  2. Phase 2

    Discovery

    2–4 weeks · fixed fee

    Paid, fixed-scope, and it ends in a plan you could hand to another firm, plus working code proving the riskiest assumption.

  3. Phase 3

    First release

    8–16 weeks

    A small senior team ships something genuinely useful into production. Real users, real data, real load.

  4. Phase 4

    Iterate & scale

    Ongoing, by agreement

    The system grows against evidence from production rather than against a plan written before anyone used it.

  5. Phase 5

    Hand-off

    2–6 weeks

    Planned from the first week, so that leaving is a scheduled event rather than a negotiation.

Engineering practice

Depth you can interrogate, not adjectives.

Nobody hires us to build a compiler or a consensus store. They hire us to build the systems that turn out to be those things with the name filed off, and we would rather be tested on that early than trusted on a brochure.

docs/adr/0014-dual-run-the-pricing-engine.mddecision record
Status    accepted 2026-03-04, supersedes ADR-0009Context   Legacy pricing is 140k lines of stored          procedures, no test suite, four known-wrong          edge cases the business depends on.Decision  Run both engines on every request for two          billing cycles. Legacy stays authoritative,          ours is shadow-scored and diffed nightly.Cut over  Nightly diff empty for ten cycles running.Cost      About 6% added write-path latency until          then. Accepted.Rejected  Big-bang cutover behind a flag. The flag was          never the risk. The absence of a comparison          was.
Written the day the decision was made, in your repository, in your name. Every engagement produces these, and the line that matters is the one recording what it cost, because a decision record with no cost on it is a press release.
  • Distributed state

    Consensus, replication, partition ownership and the reconfiguration problems that only appear when the topology changes underneath live traffic.

  • Compilers and language tooling

    Intermediate representations, dataflow analysis, and the discipline of choosing a model that makes every later stage cheap rather than merely correct.

  • Operational technology security

    Industrial protocol traffic, anomaly detection on deterministic networks, and peer-reviewed research on why enterprise tooling does not transfer.

  • Vulnerability research

    Attack-surface modelling, source review and variant analysis carried out against live third-party systems under public disclosure programs, where a claim is worth nothing until it reproduces.

  • Applied machine learning

    End-to-end pipelines where the modelling is the last and smallest step, built around evaluation sets, honest baselines and calibrated confidence.

  • Embedded and hardware interfaces

    The full path from register-transfer logic through a bus interface and a kernel driver to a user-space program, with each layer brought up and verified in order.

  • Concurrency

    Synchronization primitives implemented from the ground up, in both a forgiving runtime and an unforgiving one, until the higher-level tools stop being magic.

  • Geometry and optimization

    Computational geometry and graph search written from primitives, with benchmarking that regularly contradicts what the asymptotics suggested.

Next step

Tell us what’s breaking.

Forty-five minutes, no charge, no deck. We’ll tell you what we’d do, what it would likely cost, and whether you should be building this at all.