Five phases. Each with concrete deliverables and a reporting cadence.
Our process is transparent by construction. At every moment you know where the project stands, what decisions were made and why. We do not hide complexity behind commercial metaphors.
Execution phases
- 01
1–2 weeks
Technical briefing
We map the real use cases, the existing systems, the business and regulatory constraints.
Deliverables- A scoping document with explicit objectives and non-objectives.
- An inventory of existing systems and mandatory integrations.
- An initial risk list with proposed mitigation measures.
- 02
1–3 weeks
Architecture and estimation
We model the data domain, the access matrices, the critical flows. We define the milestone allocation and the total budget.
Deliverables- System architecture (published ADRs, database schema).
- A budget estimate per milestone, with a stated margin of error.
- A risk plan, a test plan, a release plan.
- 03
Variable, 4–16 weeks
Iterative build
We iterate in short cycles (1–2 weeks). Every release goes through staging, no exception.
Deliverables- A working demo at the end of each sprint.
- Pull requests reviewed in pairs, an integrated audit log.
- Technical documentation updated in parallel with the code.
- 04
1–2 weeks
Production launch
A controlled migration with a documented rollback plan. Smoke tests, full observability enabled from day one.
Deliverables- Runbooks for incident scenarios, handed to the client's team.
- A CI/CD pipeline with a fast rollback in case of regression.
- A technical and operational handover session with the client's team.
- 05
Ongoing, optional
Evolutionary maintenance
A written SLA with release windows, measurable monthly reporting, an evolution roadmap co-built with the client.
Deliverables- A monthly report with uptime, incidents, technical recommendations.
- Release windows agreed with the operations team.
- A roadmap revised quarterly, with business priorities.
What we don't negotiate
Direct communication
The person who signs the contract talks to the principal engineer weekly, with no intermediary.
Documented decisions
Every architecture decision ends up in an ADR. Nothing lives only in an email or on Slack.
No code in production without staging
Every release goes through staging with manual testing on the critical scenarios. No exceptions.
Rollback tested before launch
The release pipeline includes a fast recovery path, validated before the first migration.