I run an online assessment platform where users already log in, complete our tests and questionnaires, and receive detailed, individualised reports. The next step is to add an AI layer that turns those static documents into actionable, personalised learning recommendations delivered right on their dashboard. Scope of work • Build or integrate an AI engine that reads each user’s test results and questionnaire responses (both are stored in our database and tied to their account ID). • Generate short, clear learning take-aways, suggested next steps, and micro-goals unique to the individual. • Surface those recommendations as dashboard notifications the moment a new report is published, with the option for the user to refresh or request an updated set later. Technical notes – Our portal is built with standard RESTful APIs, so the AI module can run as a microservice (Python preferred, but I’m open). – You may draw on an LLM (e.g., OpenAI, Claude, or an open-source alternative) provided the solution is secure and keeps user data private. – All prompts, post-processing rules, and scoring logic need to be easy for us to tweak without redeploying code. Deliverables 1. Functional AI microservice (code + Dockerfile) integrated into our staging environment. 2. Dashboard widget or component that calls the service and displays recommendations. 3. Admin-side configuration panel where I can adjust prompt templates, length limits, and refresh frequency. 4. Brief hand-off documentation and a recorded demo walking through the flow from assessment completion to recommendation delivery. Acceptance criteria • Recommendations reference only the individual’s own results and questionnaire answers and stay under 120 words each. • Average response time from report generation to notification <5 seconds. • No PII leaks in logs; all sensitive fields must be masked or encrypted. • Code passes our security scan and unit tests you include. If this sounds like a fit, let me know your experience with similar AI integrations and any questions about our existing stack—I’ll provide API docs and sample data as soon as we connect.