Software architecture
In the AI era architecture is more than UML diagrams: when code gets cheap, structure decides.
Architecture review
- Assessment of systems that grew over time, from device to cloud
- Clear recommendations: what stays, what gets replaced, in which order
- Technical debt named and prioritised by risk and effort
Guidance for new developments
- Boundaries and interfaces between firmware, backend and app
- Security from the start, so CRA requirements do not have to be retrofitted later
- Sparring for your team during design, review and implementation
AI in the development process
- Where AI helps: speed on routine work, tests, migrations
- Where it creates debt: structural decisions nobody can oversee any more
- Guardrails for review, testing and responsibilities within the team
Reference: series-production IoT device with secure OTA
For torcbrain I developed large parts of the software of a connected cordless high-torque screwdriver: from the secure bootloader with IP protection and signed OTA updates to the device cloud that idastroem operates to this day. The product has been in the field for over two years. More in the blog.
Your contact
25 years of software development for electronics and industrial automation, from firmware to cloud operations. I design architectures I would also implement and operate myself.
Request a review