Parallel clinical workflow
22 order states with separate scan and pathology tracks, reconcile_order_status() merge layer, and package guards — ops work payment, scan, lab, and report steps concurrently.
LivoTale Platform
Healthtech · 2026
Operations hub for a liver care platform — parallel clinical workflows, CRM, and AI reporting

LivoTale is an at-home liver fibrosis screening platform. I built the admin and operations hub that ties together enquiries, orders, clinical tracks, staff onboarding, and org-wide integrations.
The dashboard gives ops and super admins filtered KPIs — enquiry funnel, conversion rate, pipeline aging, revenue, and package-wise sales — with time and dimension filters so teams can spot bottlenecks without exporting spreadsheets.
Order management is the core: a parallel clinical workflow where FibroScan and pathology run independently, with package guards for Plus, Premium, and Executive tiers. Ops step through payment, scan scheduling, lab assignment, report publish, and doctor consult — each step gated by requirements like payment received, address on file, and technician assignment.
I wired an enquiries CRM with threaded leads by phone, one-click convert-to-order, and follow-up logs. Staff HR covers technicians, doctors, lab partners, and ops teams. An integrations hub centralizes Twilio SMS, SendGrid email, OpenAI extraction, S3 storage, UPI payments, and admin-editable PDF templates.
Behind the UI: FastAPI + SQLAlchemy 2.0 async, 7 PostgreSQL schemas, transactional outbox → Redis → WebSocket notifications, and ACID order transitions with SELECT FOR UPDATE.
Technical depth
Systems and patterns behind LivoTale Admin & Ops Hub— the parts that don't show in a screenshot.
22 order states with separate scan and pathology tracks, reconcile_order_status() merge layer, and package guards — ops work payment, scan, lab, and report steps concurrently.
notification_outbox written in the same DB transaction as business logic, drained by ARQ workers to Redis, fanned out to 6+ role-specific portals.
OpenAI structured extraction from FibroScan and pathology PDFs, Jinja2 → WeasyPrint report engine with SVG health gauges and admin-editable templates.
~1,000 source files across React/Vite UI and FastAPI API, 10+ RBAC roles, GitHub Actions CI/CD, and VPS deploy with migration ledger.
Walkthrough
Key screens from LivoTale Admin & Ops Hub — what I built and why it mattered.

Enquiry funnel, revenue trend, orders over time, and enquiries-vs-orders charts — so ops can see conversion health without leaving the dashboard.

14 KPI cards — enquiries, conversion rate, pipeline aging, revenue, scan/lab/report pending counts — with today/week/month filters and package dimension filters.

Searchable patient registry with MR numbers, assigned doctors, risk scores, and status — debounced search and filter panel for ops triage.

Ops pick a date and 45-minute slot for FibroScan home visits. Booked slots grey out in real time; a requirements checklist gates the confirm button.

Payment, scan, lab, report, and consult steps run on independent tracks. Field visit progress and FibroScan verification sit below scheduling so ops see the full pipeline.

Billing/GST, UPI payments, Twilio SMS, SendGrid email, OpenAI extraction, and S3 storage — each with configured/not-configured status and admin-editable templates.

Per-team KPIs for new enquiries, conversion rate, orders created, unpaid orders, and revenue — with enquiry funnel and orders-over-time charts.

Order detail Reports tab with AI-hybrid liver health PDF — FibroScan LSM/CAP, FIB-4, APRI, FAST scores, and interactive vs PDF view toggle.

Org-wide login attempts with success/failure status, IP addresses, and failure reasons — full audit trail for staff access across all roles.
LivoTale Platform