BI / Dashboard Developer (Claude AI and/or Python + Power BI) for UK aesthetics clinics (ongoing)

Long-term BI partner wanted: turn our working AI-built dashboards into a secure, self-serve reporting portal for a UK healthcare group. Who we are We run a group of aesthetics/medical clinics in the UK across three brands, ~11 locations and ~40 practitioners. The three brands run on three different clinic management systems - Zenoti, Pabau and Timely - which is exactly why reporting is hard, and why this role exists. Working with Claude (Anthropic's AI), we have built a suite of five interactive KPI dashboards that we use ad hoc today: one per brand, a consolidated group view with forecasts, and a per-practitioner performance dashboard. They are self-contained HTML/JavaScript pages fed by Python data pipelines that ingest raw exports from the three booking systems, apply a carefully verified methodology (VAT treatment, new-patient definitions, cross-system reconciliation to our management accounts) and refuse to publish if any of ~25 automated checks fail. The numbers need to be trusted; the methodology is documented. You will inherit all of this code and documentation. What they are not yet: automated, multi-user, or connected live to source systems. We also need them to be totally secure to protect client data. Today, refreshing them means exporting files and running the pipeline - and that makes one person the bottleneck. What we need you to build - Phase 1 (the core project) Automated data refresh. Replace manual file exports with scheduled pulls from Zenoti, Pabau and Timely (APIs where available, scheduled report exports where not). Data lands in a small, secure data store; pipelines run on a schedule; dashboards are always current when the team logs in on Monday morning. A simple, secure portal. One URL where each team member logs in and sees the dashboards for THEIR role — full group view for directors, clinic view for clinic managers, their own performance page for each doctor, marketing views for the marketing team. Viewing must be one click; only 2–3 admins can change anything. Cost-safe architecture. Dashboards must be pre-computed on refresh so that viewing costs nothing - no AI/API credits, no expensive compute per page-view. AI (Claude) is a build/maintenance tool for you, not something 30 staff burn credits on daily. You will propose the stack (we are open: Power BI with proper workspaces/RLS, or a lightweight web portal serving our existing HTML dashboards but depends on what fits). Security, closed-loop. This is health-adjacent data with patient names. Everything private by default, UK/EU data residency, least-privilege credentials, no third-party analytics leaking data, GDPR-compliant. You'll sign an NDA and a data processing agreement, and we'll expect you to work with pseudonymised data wherever possible. Phase 2 (shortly after) New data sources for the lead funnel: front-of-house phone call data (call tracking / phone system logs) and Google Analytics 4, joined to the booking systems so we can finally see lead source → enquiry → booking → conversion → patient value, and judge our PPC spend properly. Role-specific dashboard variants built out fully (marketing, front-of-house, doctors, clinic managers) on the shared data model. Then: ideally a long-term relationship Once built, this becomes a part-time retainer: monthly maintenance, small changes, new metrics and occasional new dashboards as the business evolves. We want one person who knows our stack deeply, responds quickly, and grows with us. Who we're looking for Proven Claude & Power BI (or equivalent production BI) experience - workspaces, row-level security, scheduled refresh, real deployments used by real teams. Show us examples. Strong Python / ETL skills and experience building API connectors - bonus points for clinic management systems, CRMs, GA4 or call-tracking platforms. Comfortable doing AI-assisted development with Claude (Claude Code or similar) - our existing codebase was built this way and we work this way. You don't need years of it (nobody has that), but you must be fluent and enthusiastic. Excellent design/UX instinct. Our dashboards get shown to boards and doctors; they must be beautiful and instantly readable, not default-template Power BI grey. Detail-obsessed. Our current pipelines reconcile to management accounts and refuse to publish when checks fail. If "the numbers must tie" excites rather than annoys you, you'll fit right in. Commercially minded. You should understand why a clinic cares about ATV, rebooking, utilisation and new-patient conversion - and push back when a metric is misleading. Responsive, fast-moving, with at least 2–3 hours of working-day overlap with UK time. Engagement & budget Phase 1: fixed-price, milestone-based (milestones: architecture proposal → automated refresh working for one brand → all three brands → portal with role-based access → handover docs). State your price and timeline in your proposal. Ongoing: part-time hourly retainer (expect a few hours most weeks, more when we're building something new). How to apply Start your proposal with the word AESTHETIC. Link 2–3 dashboards or data products you personally built, and say which parts were yours. In 5–8 sentences: how would you architect Phase 1? (Stack, where the data lives, how refresh runs, how role-based viewing works, how you'd keep running costs near zero.) One paragraph: an example of a data-accuracy problem you caught that others missed. Your Phase 1 fixed price, your hourly rate for the retainer, and your availability/timezone. Generic proposals and obvious AI-written applications without the keyword will be declined. The right person gets a long-term, respectful client who pays on time and knows exactly what they want.