Shaping a technical concept into a complete product system
Client
DeepEdge · via Wolffkraft
Service
Product architecture · UX design · Machine learning & edge AI
Engagement
6+ months · Early concept to MVP

DeepEdge brings data preparation, model training, optimisation and deployment into one platform for building AI applications that run on edge devices.
Over 6+ months, I worked with the engineering team and a visual designer to translate an early technical concept into a usable product.
My contribution covered product architecture and workflows for people with different responsibilities and levels of technical expertise—from developers working with models to annotators preparing data and administrators managing the platform.
When I joined, engineers were carrying out operations through the terminal. The technical foundations were taking shape; the challenge was to turn them into workflows that people beyond the engineering team could understand and use without losing the technical depth specialists needed.
Defining how people would work with the system
Establishing the product structure
Mapped the system’s entities, relationships and behaviours to define how its capabilities would fit together.
Supporting different responsibilities
Shaped workflows around the needs of developers, annotators, administrators and viewers.
Making complexity manageable
Questioned configuration requirements and structured longer processes so users could follow their progress.
01 - PRODUCT ARCHITECTURE
Turning technical logic into connected workflows
Understanding the system with engineering
Worked closely with a senior engineer to map the operations behind the platform and the relationships between its parts.
Defining behaviour around users
Examined how each entity should behave depending on who was interacting with it and what they needed to accomplish.
Giving the product a shared foundation
Used this mapping to connect technical requirements with user workflows, creating a basis for detailed design as the platform evolved.
02 - COMPLEX WORKFLOWS
Making demanding processes easier to follow
Questioning the setup burden
Worked through the purpose of each flow with engineering to identify which configurations users actually needed to make.
Structuring necessary complexity
Organised longer workflows into understandable steps while retaining the detail needed for technical work.
Keeping users oriented
Made progress visible so people could understand where they were in the process and what remained.
Bringing a product perspective to an engineering-led platform
01
Finding clarity in an unfamiliar domain
Built enough understanding of machine-learning operations to question assumptions and contribute to product decisions.
02
Creating a shared language with engineering
Used maps, flows and structured documentation to make emerging ideas concrete, discuss gaps and agree on how the product should behave.
03
Establishing a foundation for growth
Developed reusable interaction patterns with a visual designer, giving the MVP a consistent basis for capabilities still taking shape.






