Nexiv Labs is now operating six divisions across healthcare, veterinary, commerce and infrastructure. See the group →
Division 04 — Product engineering

Nexiv Technology

The engineering division of the group. Full-stack web, mobile, data platforms and applied AI — delivered by embedded squads that have actually shipped software into regulated markets.

Scope an engagement The stack
Trunk-based delivery Security in the pipeline Embedded squads
Commit → production
Disciplines

Six practices, one delivery team

We do not hand a project between departments. A squad carries product, design, engineering and quality together, from first discovery call to production support.

Web & product engineering

TypeScript end to end. React and Next.js on the front, Node/NestJS or Go behind it, typed contracts between them, and a design system that keeps twenty screens looking like one product.

TypeScript Next.js NestJS

Mobile

React Native where a shared codebase wins, native Swift and Kotlin where the platform demands it — with offline-first sync, biometric auth and store-release automation.

React Native Swift Kotlin

Data platforms

Event streaming, warehouse modelling and analytics that people actually trust — because lineage, freshness and test coverage are visible next to every number.

Kafka dbt PostgreSQL

Applied AI

Retrieval over your own corpus, tool-using agents with hard guardrails, and an evaluation harness that runs on every change so quality is measured rather than asserted.

RAG Evals Guardrails

Quality & assurance

Unit, contract, integration and end-to-end tests in the pipeline, plus load testing, accessibility auditing and security review before anything reaches a real user.

Playwright k6 SAST/DAST

Modernisation

Legacy systems replaced incrementally with the strangler-fig pattern — routing traffic slice by slice, so there is never a single terrifying cutover weekend.

Strangler fig Anti-corruption layer
The stack

What we reach for, and why

Defaults, not dogma. We choose boring, well-understood technology and spend the novelty budget on the actual product problem.

Front end
  • FrameworkNext.js with the app router; server components where they reduce payload, client components where interaction demands it.
  • LanguageTypeScript in strict mode. No implicit any, no unchecked API responses.
  • StateServer state via query caching; client state kept deliberately small and local.
  • StylingDesign tokens driving a component library; dark and light parity from day one.
  • AccessibilityWCAG 2.2 AA target; keyboard paths and screen-reader labels reviewed per component.
  • PerformanceCore Web Vitals budgeted in CI; a regression fails the build rather than the launch.
Back end
  • ServicesNode/NestJS for product surfaces, Go for throughput-sensitive paths, Python/FastAPI for ML serving.
  • ContractsOpenAPI and protobuf generated from source; clients generated, never hand-written.
  • MessagingKafka for domain events, outbox pattern for transactional publishing.
  • AuthOIDC, short-lived tokens, mTLS between services, authorisation decisions centralised.
  • RuntimeContainers on Kubernetes, managed by Nexiv Cloud with the same SLOs as our own products.
Data
  • OperationalPostgreSQL with row-level security; migrations reviewed and reversible.
  • CacheRedis for sessions, rate limiting and hot reads with explicit invalidation rules.
  • AnalyticsWarehouse modelling with dbt; every metric has an owner, a test and a definition.
  • GovernanceColumn-level classification, retention policy per dataset, PII minimised at ingestion.
  • LineageSource-to-dashboard lineage published so a number can always be traced back.
Applied AI
  • RetrievalHybrid keyword + vector retrieval with chunking tuned per corpus and cited sources in output.
  • EvaluationGolden datasets and regression evals in CI; a quality drop blocks the release.
  • GuardrailsInput and output filtering, tool allow-lists, and hard refusal paths in clinical contexts.
  • Human oversightIn regulated flows the model drafts and a qualified human decides — never the reverse.
  • GovernanceModel, prompt and dataset versions recorded per inference for auditability.
Quality & security
  • TestingUnit, contract, integration and Playwright end-to-end suites gate every merge.
  • Loadk6 scenarios modelled on real traffic shapes, run before every major release.
  • Supply chainSBOM generated per build, dependencies scanned, artefacts signed and verified at deploy.
  • Application securitySAST and DAST in the pipeline; threat modelling at design time for new surfaces.
  • AccessibilityAutomated axe checks plus manual keyboard and screen-reader passes each release.
Engagement

Three ways to work with us

Whichever shape you pick, you get the same squad model and the same delivery standards — the difference is who holds the roadmap.

Outcome delivery

We own scope, timeline and quality against an agreed outcome. Best when the destination is clear and you want a single accountable team.

Embedded squad

A cross-functional team joins your organisation, works in your rituals and your repositories, and reports into your product lead.

Technical due diligence

A short, blunt assessment of an existing codebase, platform or team — architecture, security posture, delivery capability and real risk.

Compare engagement models
.github/workflows/delivery.yml 
# Every merge to trunk runs the same gate.
jobs:
  verify:
    steps:
      - run: pnpm typecheck        # strict TS, zero errors
      - run: pnpm test --coverage  # unit + contract
      - run: pnpm e2e --shard       # Playwright, parallel
      - run: pnpm axe:ci            # accessibility
      - run: syft . -o spdx-json    # SBOM
      - run: grype sbom:. --fail-on high
      - run: cosign sign $IMAGE     # signed artefact

  deploy:
    needs: verify
    strategy: canary              # 5% → 25% → 100%
    rollback: automatic-on-slo-breach
Deploys / week
40+
Lead time to prod
< 1 day
Questions

Common questions

You do. Work product created for a client engagement transfers on payment, as set out in the Terms of Service. We retain rights only in pre-existing tooling and generic libraries we bring with us, and you receive a perpetual licence to use those as part of the deliverable.
Yes — that is the default for embedded squads. We work in your version control, your CI, your ticket tracker and your review process. Where your pipeline lacks a control we consider essential, we will propose adding it rather than working around it.
Handover is designed in from the start, not bolted on at the end: architecture decision records, runbooks, an onboarding guide and a pairing period with your team. We would rather leave a team that no longer needs us than a dependency.
For well-bounded scopes, yes. For discovery-heavy product work we prefer a capped time-and-materials model with a firm review gate, because fixed price on an unclear scope usually means either padded estimates or a fight about change requests. We will tell you honestly which category your project is in.

Tell us what needs building

A brief, a broken system, or just a hypothesis — we will tell you what we would do and roughly what it costs before anyone signs anything.

Scope an engagement See the platform