Thriveworks
Major contributorTechnologies
Languages
Frontend & UI
Backend & Database
AI & Integrations
Testing & Analytics
Infrastructure & Tools
Thriveworks is a large-scale mental health therapy platform serving patients across the United States. The platform enables patients to search for therapists, book appointments, manage insurance, and access care through a modern web application backed by a robust API layer.
My Contribution:
One of the earliest engineers on this project, contributing for ~5 years as a full-stack engineer from day one (not a frontend role that later expanded into backend). Built core patient-facing features on the Next.js frontend (provider search, booking flows, and early-phase insurance-related work only — not end-to-end insurance handling) and on the NestJS API, including architecting the AI module and real-time voice pipeline (voice is not live in production yet). Contributed to test infrastructure.
Provider Search & Discovery
Built and iterated on the provider search engine — the primary patient-facing feature for finding therapists.
- Advanced filtering: location, insurance, specialty, counseling type, availability
- Timezone handling for provider availability and discovery
- Sort/filter UI with mobile-responsive filter modals
- Provider profile pages with SEO-friendly structured data (JSON-LD)
Booking & Appointment System
Contributed to the multi-step booking flow end-to-end for the patient journey — one of the most complex user journeys in the app. Full insurance verification and downstream insurance handling were out of scope; work on insurance stayed in an initial phase only.
- New booking flow layout with consent management, billing integration, and patient form handling
- Support for individual, couples, family, and medication management counseling types
- Cross-state booking handling with blocker components
- Cancellation policy reminders and session-within-48-hours hooks
AI & Voice Integration (in development)
Implemented AI-powered natural language search using OpenAI so patients can describe their needs conversationally. This is a full search experience wired as a system: an AI search agent on the frontend is connected to the NestJS backend via Socket.IO, which drives filters, search results, and URLs in real time from the conversation — not a separate built-in chatbot widget. Real-time voice and the full AI/voice rollout are not released yet; the backend AI module is architected to support that path.
- MCP (Model Context Protocol) tools for location and provider search
- Socket.IO between the AI search agent and NestJS backend for live UI and search updates, with gateway work for the real-time layer (voice path still in development)
- Dual-agent architecture with asynchronous tool execution
- Vercel AI SDK for model streaming and text-to-speech with progressive audio playback
Internal Testimonials Tool
Built an internal testimonials management tool from scratch — a dashboard for operations teams to bulk upload, update, and manage user testimonials across the platform. Implemented CSV bulk upload, CRUD interfaces with rating validation, and service/specialty filters.
Other Contributions
- Auth0 integration with protected route handling in the useApi hook
- Provider agenda optimization: batched fetching to resolve N+1 query issues across license keys
- Coordinated and individually built a feature for the company hackathon