Mission MNT

The Problem

Mission MNT combines metabolic medication, clinical consultations, nutrition, and 1:1 coaching into one program. When I came on board, they needed to turn that vision into a working digital clinic.

The biggest challenge wasn’t just building an attractive patient dashboard. Healthtech has weird, asynchronous constraints: a user can’t just pay and get their medication 10 seconds later. A licensed prescriber has to manually review their medical questionnaire, verify their identity, and approve the prescription first. If they’re ineligible, they need a refund. If they’re approved, an external pharmacy has to dispense and ship it.

I worked as the sole UI/UX engineer bridging product design and full-stack development, responsible for mapping the user journey, designing the interfaces in Figma, and writing the production code.

Rapid Prototyping & Exploring Complex Flows with AI

Because I was handling both product design and engineering across a deeply regulated flow, velocity and edge-case discovery were everything. I used AI as a force multiplier throughout the design and prototyping phases:

• Stress-testing state permutations: Healthcare flows have dozens of edge cases (e.g. initial algorithmic pass → checkout → clinician query → identity verification failure → refund). I used AI tooling to map out decision trees and uncover hidden failure states before touching Figma or the database schema.
• Rapid code & UI prototyping: Instead of spending days wireframing abstract static screens, I used AI alongside rapid Next.js scaffolding to spin up working, interactive prototypes of complex flows—like the post-checkout "Waiting Room" and the clinician cohort triage view. This let us test the tactile feel of loading states, banner messaging, and transitions with our clinical team and leadership within hours rather than weeks.
• Exploring API integration models: When negotiating the integration with PHLO, I used AI to quickly generate realistic mock webhook payloads and test how our state machine would react to unordered delivery, duplicate events, and error scenarios.

Designing for Asynchronous Care: The "Waiting Room"

Most e-commerce checkouts are instant, but healthcare isn't a simple binary switch. Between a patient paying and a clinician signing off, there’s an unavoidable limbo period.

If you just lock the user out with a "We're reviewing your application" screen, they get anxious and immediately email support asking where their money went. If you give them full access immediately, they might book a coaching call before we even know if they’re medically allowed to be on the medication.

To solve this, I designed and built a staged "Waiting Room" flow:

• Upfront algorithmic screening: We filtered out hard medical contraindications before account creation or payment, so ineligible users weren't charged in the first place.
• Immediate post-payment access: The moment someone paid, their account was active, but feature-gated. They entered a tailored waiting room dashboard where they could track their prescription review status in real time, upload baseline body metrics, and browse onboarding guides.
• Silent gating: High-touch features like booking 1:1 coaching calls or managing prescription protocols were disabled until the clinician's approval webhook landed.
• Clinical rejections: If a prescriber rejected an application, the UI didn't just crash or show a generic error. The dashboard cleanly transitioned into an empathetic rejection state explaining why, while our backend kicked off an automated cancellation and refund.
• Churned members: When someone cancelled their subscription, we didn't lock them out of the platform. They kept access to their historical weight graphs and past check-ins, but coaching was disabled and the dashboard surfaced an easy re-consultation flow so they could restart treatment down the line without starting from scratch.

Deep API Integrations & Working with PHLO's CTO

Our prescription engine and fulfillment ran on PHLO, a major UK digital pharmacy. I worked directly with PHLO’s CTO and their engineering team to iron out the technical contracts between our platforms.

Clerk served as our identity provider, which PHLO trusted via OIDC. But the real engineering complexity was in the webhook lifecycle:

• Out-of-order events: In the real world, webhooks don't arrive in chronological order. Because pharmacy fulfillment happens across different microservices, we ran into race conditions where a medication "dispensed" event could hit our webhook endpoint before the clinician's "approved" event finished processing. I built our ingestion handler to sequence updates using event timestamps (occurredAt) rather than arrival time, preventing state from rolling backwards.
• Idempotency & retries: PHLO uses an at-least-once delivery model that retries failed deliveries for 24 hours. I implemented an idempotency key layer backed by Postgres so duplicate retries were safely acknowledged without triggering duplicate database writes or customer emails.
• Marketing CRM vs. Medical Privacy: I worked closely with our Director of Marketing to sync lead stages and user activity into HubSpot for abandoned checkout remarketing. However, because this is medical data, I had to ensure strict data boundaries—questionnaire answers, specific medical flags, and clinical notes were strictly isolated to our HIPAA/GDPR-compliant database and never forwarded to marketing tools.

Clinical & Admin Operations

Patients were only half the equation. Our clinical and coaching staff needed a way to manage the patient population daily.

I designed and engineered an internal Admin Dashboard with:

• Cohort filtering: Quick views for clinicians to triage members by their live status (pending_review, active, requires_attention).
• Role-Based Access (RBAC): Using Clerk metadata and Postgres Row Level Security, I separated permissions cleanly. Coaches could review weight trends, nutrition habits, and check-in notes, but had zero access to private prescriber logs or raw medical questionnaires.
• Attention triggers: Automated flags that surfaced patients who missed required monthly check-ins or reported adverse side effects during dose titration.

What I Took Away

This project was a great balance of systems-level thinking, executive communication, and granular frontend/backend engineering.

Because I was in the room negotiating API specs with PHLO’s technical leadership, aligning marketing lifecycle logic with our growth director, and writing both the frontend components and backend webhook handlers, there was zero translation loss between how the business wanted the clinic to run and how it was actually coded. Using AI to compress the discovery and prototyping cycles gave me the leverage of a small team while maintaining total ownership over the end-to-end craft.