If it doesn't reproduce,
it isn't a result.
Science runs on reproducibility — a result that can't be reproduced isn't a finding, it's a guess. Too much research compute is a tangle of scripts that give a different answer on a different machine. We serve research two ways : deterministic compute on Heisen — reproducible analysis, HPC verification, data provenance, FAIR-aligned records — and Murlink precision rigging for instrument + detector handling. Same input, same result, provenance intact. The lab owns the science ; we make the compute behind it reproducible.
Bit-exact
Same input, same result, every run — deterministic
FAIR
Findable, Accessible, Interoperable, Reusable data
Provenance
Every result traceable to inputs + code version
Sovereign
Canada-built, data stays in your control
Six exposures across the research compute.
Research risk is reproducibility + provenance risk : a result that won't replicate, a dataset with no traceable lineage, an instrument move that damages a one-of-a-kind detector. The exposures below are what deterministic, audit-grade software + precision rigging exist to control.
Non-reproducible result
A result that gives a different number on re-run or on another machine can't be published, peer-reviewed, or built upon. Heisen is deterministic — same input, same result, with inputs + code version pinned for replay.
Lost data provenance
Data with no traceable lineage from raw to result can't be defended at review. Provenance is captured automatically — every result traceable to its inputs, transforms, and code version.
Numerical drift across HPC
Floating-point results that drift between nodes or library versions silently corrupt a long computation. Deterministic compute pins the numerics — the result is the result, regardless of where it ran.
FAIR non-compliance
Funders increasingly mandate FAIR data. A dataset that isn't Findable, Accessible, Interoperable, Reusable is a grant + reuse problem. We build records to the FAIR principles.
Precision-instrument damage
Moving a detector, magnet, or optical assembly with steel rigging risks contamination + damage to a one-of-a-kind instrument. Murlink synthetic rigging is non-marking + precise — the handling that protects the instrument.
Data sovereignty
Research data on a foreign-controlled platform is a sovereignty + IP exposure. Heisen is Canada-built + sovereign — the data + the compute stay in the institution's control.
Software-led, with precision handling.
Research is a software engagement with a precision-rigging tail — deterministic compute, provenance, HPC verification, and Murlink for instrument moves. Six concrete fits.
Deterministic analysis pipelines
Reproducible analysis — every transform recorded, version-pinned, replayable. The result that holds up to peer review and builds into the next study.
HPC verification + reproducibility
Deterministic verification across HPC — numerics pinned so a long computation gives the same result on any node, any run.
Data provenance + lineage
Automatic provenance capture — every result traceable to its raw inputs, transforms, and code version. The lineage a reviewer or funder expects.
FAIR-aligned data records
Records built to the FAIR principles — findable, accessible, interoperable, reusable — meeting funder mandates and enabling reuse.
Precision instrument + detector rigging
Dyneema® synthetic rigging for moving detectors, magnets, and optical assemblies — non-marking, precise, non-conductive. The handling that protects a one-of-a-kind instrument. Authorized Quebec distribution.
Embeds into your research stack
Heisen embeds by API into your existing HPC / data-management stack — the deterministic + provenance layer alongside your tools, data residency intact.
Scientific Research in the field
From sector context to the lifts we engineer — a look at where this work happens.
The standards we work to.
We build the software to the reproducibility + data standards your research is held to, and supply rigging to the crane-rigging standard. The science is the researcher's — we build the deterministic compute + provenance around it.
FAIR data principles
Findable, Accessible, Interoperable, Reusable — the data principles funders increasingly mandate. Our records are built to FAIR from the start.
Models + simulations credibility
NASA's standard for the credibility of models + simulations — verification, validation, reproducibility. The discipline our deterministic compute satisfies.
Testing + calibration laboratories
The competence standard for testing + calibration labs. Our data + record software aligns to the ISO 17025 framework your lab is accredited against.
Data-integrity principles
Attributable, Legible, Contemporaneous, Original, Accurate + Complete, Consistent, Enduring, Available. The research record holds to ALCOA+ for review + audit.
Slings + rigging
Governs slings + rigging. Cited on the Murlink rigging spec for precision instrument + detector handling — each component rated + contamination-selected.
Information security management
The ISMS standard. Our software keeps research data + IP secure and resident under an ISO 27001-aligned posture.
Software-led. Murlink for precision handling.
Research is software-led with a Murlink tail. Our crane-lift pillars step back ; the value is the deterministic compute, with precision rigging for instrument moves. Here's how each pillar applies, honestly, including where it doesn't.
Sealed plans + emergency response
Not applicableHeavy crane lifts don't apply to a research-software engagement — there's no crane pick in our scope here. The engagement is the deterministic compute and the precision rigging.
Distribution + training + implementation
Not applicableWith no crane lift in scope, CRANEbee lift simulation doesn't apply to this sector. Our modeling value is the reproducible research compute, not a crane study.
Distribution + advisory + training
Dyneema® synthetic rigging for moving detectors, magnets, and optical assemblies — non-marking, precise, non-conductive, no contamination on a one-of-a-kind instrument. Authorized Quebec distribution, with advisory on the spec.
Deterministic engineering platform
The lead. Deterministic research compute on Heisen — reproducible analysis, HPC verification, data provenance, FAIR-aligned records — audit-trailed, plus Maxor Audit + Maxor Ground. Sovereign, Canada-built, data residency intact. Heisen embeds into your HPC / data stack by API.
Research compute that reproduces, years later.
For research, the engagement is the software — anchored on Heisen. We build deterministic, audit-grade compute : same engineering posture, same team kickoff to go-live, sovereign by default. An analysis that returns the same result on re-run years later, with the provenance to prove exactly how the result was produced. Engineered in Canada, owned by your institution.
Heisen — our deterministic intelligence layer — is optional on any build: embed it or not, your call. Either way it plugs into a fresh custom app or your existing third-party software via API.
Deterministic analysis + HPC platform
Reproducible analysis pipelines with pinned numerics — the same result on any node, any run, version-pinned for replay. Not a script tangle that drifts between machines.
Provenance + FAIR data platform
Automatic provenance capture + FAIR-aligned records — every result traceable to its lineage, meeting funder mandates and enabling reuse, sovereign and resident.
Verification + reproducibility record
An audit-trailed verification record — every computation reproducible and defensible at peer review, the credibility the science depends on.
Scope your research engagement.
Tell us the program, the compute, and the reproducibility + data obligation. A senior lead responds within one business day with a scoped engagement and a path to first deliverable.