I worked on redesigning this learning platform while at DSSL / TRASSIR. Here I revisit that real project with my current approach: start with the specialist’s task, then connect a useful technical answer to a structured learning path.
Outcome. A technical answer can be read before sign-in and reused inside a learning programme. Reading, completed units and assessment results remain separate facts. What it cost: explicit completion, staged access and a smaller qualification promise than a conventional training portal makes.
The archived product opens with a course catalogue: search, topic and brand filters, ratings and rewards. Another original screen asks for company details before unlocking courses. These screens show the starting structure; they do not tell us how well real users completed their work.
For the redesign, the useful distinction was between two situations. An engineer investigating a camera that cannot be discovered needs a sequence of checks. A designer asked to learn the product range needs a route, an estimate of the work and a clear finishing condition. Sending both into the same course list leaves them to reconstruct that route themselves.
The central hypothesis was that reference and learning could share content without sharing every rule. A material has its own address, product version and update date. It can answer one question independently and occupy a named place inside a programme. The reader should not have to enrol simply to discover whether it is relevant.
That also changes the meaning of progress. Opening a troubleshooting article is evidence of reading, not evidence that its learning objective has been completed. The two records need to meet in one personal space while remaining independent. Otherwise an innocent page view starts making qualification claims.
Give the answer its own address. Let a programme organise those answers without turning every visit into a completed lesson.
For this rework, I returned to the original screens and the problem I had worked on in the company. I added secondary research and competitor walkthroughs. The personas and interviews used in this new iteration were synthetic: useful for challenging assumptions, insufficient for claiming demand. They informed two equally important acceptance routes — an engineer finding a technical answer and a designer starting and completing structured training.
The PRD then fixed the access ladder, content model and counting rules before the sitemap, design system and interactive build. The public programme page exposes its purpose, workload, full contents and assessment conditions before sign-in. Both the service and its landing are available in English and Russian; links from this portfolio open the selected language.
The final browser acceptance found a primary button whose label inherited a muted text colour. The action existed and the screen was assembled, but the next step was difficult to read. The fix belonged in the shared text rule, followed by checks of sign-in, explicit completion and next-unit navigation at three widths.
This new version still needs subject-matter review of its question bank. Some questions rely on sources outside the selected programme. Those sources are disclosed before assessment, but disclosure cannot establish that the assessment measures competence. The prototype demonstrates my current design approach; it does not establish learning outcomes for either this version or the original work project.
Four content colours carry the task from home into the catalogue, material and player. Status colours have a separate meaning. Assessment deliberately switches to neutral, while version and update information stays neutral everywhere. There are no points, streaks or success celebrations substituting for progress.
The landing shares that visual vocabulary but uses its own type scale, spacing and component set. It proves the promise with a material and a programme rather than invented testimonials. Its first useful action opens the product; English and Russian lead to the corresponding product language.
These engineering checks apply to the current portfolio version. They do not describe testing or measured results from the original work project. This version has not been released as the company’s product. Screen-reader and physical-device testing remain open; storage lasts within the browser tab, messages and company checks are simulated, and a full offline application is not implemented.
That boundary shaped the content, access, progress model and recovery behaviour throughout the product.
If the case above answered your question — or raised one.
