Engineering built to last.
Cloud, data and product engineering for systems that need to scale.
The firm
Twelve engineers, one office, long engagements.
“Most of what we are asked to fix was built correctly for a company half the size.” Engineering practice, Visionmost
Visionmost is an engineering firm based in Nicosia, Cyprus. We work on systems that already carry real load, complexity and responsibility.
Our engagements combine assessment, embedded engineering and handover — so the client’s team can confidently run what we build.
Practices
Four practices, and the work that actually arrives in them.
Platform engineering
Infrastructure described in code, environments that match each other, and a deployment path a new engineer can follow on their first week without supervision.
- Kubernetes and container platform design
- Terraform modules and environment parity
- Release pipelines with gated rollout and rollback
- Observability: metrics, traces, and alerts that page a human for a reason
- Cost review before and after migration
Data infrastructure
Getting records in reliably, storing them in a shape that answers the questions the business actually asks, and being able to show where any number came from.
- Batch and streaming ingestion
- Warehouse and lakehouse modelling
- Lineage, quality checks and reconciliation reporting
- Access control and retention aligned to GDPR obligations
- Migration from legacy reporting stacks, in stages
Applied machine learning
We work on models that have to run every day rather than demonstrate well once: forecasting, classification, document extraction, anomaly detection in operational data.
- Feature pipelines and training infrastructure
- Evaluation harnesses and regression testing on model output
- Serving, versioning and rollback
- Drift monitoring with a defined human review step
- Documentation for internal risk and compliance review
Product engineering
The applications people open to do their work: operator consoles, customer portals, and the mobile clients that go into the field with them.
- Web applications in TypeScript and React
- Mobile clients in Swift and Kotlin
- Interface design and a maintained component library
- Accessibility review against WCAG 2.2 AA
- Front-end performance budgets held in the pipeline
Capability
What we assess, build and hand over in each domain.
| Domain | We assess | We build with | We hand over |
|---|---|---|---|
| Cloud platform | Environment drift, deployment risk, cost profile | AWS, Google Cloud, Kubernetes, Docker, Terraform | Infrastructure repositories and runbooks |
| Data | Lineage gaps, freshness, reconciliation failures | PostgreSQL, BigQuery, Kafka, dbt, Airflow | Documented models and quality checks |
| Machine learning | Evaluation coverage, drift exposure, review process | Python, PyTorch, MLflow, feature stores | Training pipelines and evaluation reports |
| Applications | Accessibility, performance budget, dependency risk | TypeScript, React, Node.js, Swift, Kotlin | Component library and test suites |
Method
How an engagement runs.
Assessment first, always
Every engagement opens with three to four weeks of assessment: reading the system, talking to the people who operate it, and writing a document that states what we found, what we would change, and in what order. It is delivered whether or not you continue, and it is written to be readable by a non-engineer on your board.
Embedded, not adjacent
Our engineers work in your repositories, your issue tracker and your standups. There is no parallel Visionmost backlog and no separate status report. If your team cannot see what we changed on the day we changed it, we have set the engagement up wrongly.
Exit is part of the plan
Each engagement names the conditions under which it ends: capabilities transferred, documentation accepted, on-call rotation staffed by your side. We schedule the handover at the start rather than negotiating it at the end, and we reduce our involvement in stages so the transfer is tested before it is final.