Selected work

Things we have built and run

Client names stay private. Ask on a call and we will go into detail.

Agentic delivery · Program

An agentic software development lifecycle, from assessment to first agent-built release

A large customer-operations company wanted AI agents to carry routine development work, with controls its leadership could defend.

The brief

Find out how ready the teams and the code were, define a lifecycle where agents build and people decide, and prove it on real work rather than a demo.

Outcome

Agent-built changes merged under human review on production repositories, and a roadmap for scaling that the client owns.

What we did

  • Readiness assessment of teams, repositories and practices; a scope that matched the actual maturity rather than the wish list
  • The lifecycle and its gates: intake and planning, agent dispatch within limits, human review, merge, deployment and evaluation, with evidence kept for every change
  • A first increment delivered through it with the client's own people, a playbook for the next teams, and a coaching rhythm to keep it honest
AI · Platform

An agentic AI framework for a software company

A software company wanted its own foundation for AI agents rather than a pile of one-off scripts.

The brief

Build the framework its product teams would extend: agent services, orchestration, data and the infrastructure around them, ready to split into services as it grows.

Outcome

A working foundation the client's teams extend on their own, delivered on their platform, in their accounts.

What we did

  • An agent service layer with a clear contract, an orchestration interface to run and observe agents, and a data platform behind them
  • Infrastructure as code and deployment pipelines for the whole thing, in one repository structured to be split later
  • Documentation and examples so the client's developers could add agents without us
SaaS · Cloud

A data platform for a scientific-instrument startup

The founders wanted their customers to make sense of the data their instruments produce. They had beta testers waiting and no software team.

The brief

Design and build a web platform that turns raw measurement files into shareable, interactive reports, from the first commit to the first beta users.

Outcome

A product in front of beta testers within the first quarter, and environments the founders understand and own.

What we did

  • Product and architecture from scratch: back end, front end, database, authentication, encrypted data
  • Staging and production on the client's cloud, with monitoring, error tracking and a running-cost estimate per environment
  • Sprints against a fixed timeline to the first release, then iteration on tester feedback; staging later moved to private infrastructure to cut the cloud bill to zero
Medical devices · Leadership

Medical-grade real-time software under FDA and IEC 62304

Years of leading engineering teams at medical-device companies, where every release has to satisfy regulators and still ship on time.

The brief

Deliver real-time, imaging and 3D software for medical devices across teams in several countries, under full regulatory scrutiny.

Outcome

Features delivered on time with audit-ready documentation, and teams that kept growing. This is the discipline Inellova brings to smaller companies.

What we did

  • Architecture, data structures and algorithms for real-time processing; releases run end to end with the documentation auditors expect
  • Services moved to containers and orchestration on the cloud, with observability introduced along the way
  • R&D in computer vision, 3D imaging and language-model features, taken from experiment to product
Smaller jobs

Not everything is a big project

Smaller pieces of work we also take on, at a fixed price.

Cloud foundationsLanding zones, identity, networking and policies, written as infrastructure-as-code modules and templates your team reuses
PipelinesBuild, test and release pipelines with environments, approvals and rollback; self-hosted build agents where the cloud ones won't do
MigrationsSource control, work tracking and pipelines moved between platforms; legacy applications moved to the cloud, one wave at a time
Cost and governanceA cloud bill explained line by line, cut where it should be, and guarded by budgets, tags and policies
Security hygieneSecrets out of code, least-privilege access, dependency and vulnerability scanning, a backup you have actually restored
AI, where it earns its keepA coding assistant rolled out with rules, a document search over your own files, an agent that does one boring job well

Ready when you are

Thirty minutes, free. You will know quickly whether it fits.

Book a free call