Full-Stack / Video Pipeline Developer – Automated AI/Dynamic Trip Highlights Reel for Travel App

Job Type Fixed-Price (Milestone-based) or Hourly (Expert level) | Estimated time: 1-2 weeks 1. Project Overview & Context GoPilgrim (gopilgrim.lovable.app) is a mobile-first web app built for youth leaders, churches, and mission trips to keep students safe and families connected in real-time. Currently, leaders track GPS checkpoints, take roster attendance, and share audio updates. We are now building our signature parent delight feature: Trip Story / Daily Highlights Reel. The Vision: Leaders snap photos and short video clips during the day. In the evening, the system automatically curates, trims, and renders a branded 30–60 second daily highlight video (complete with background audio, trip stats, and day captions) for leaders to review, approve, and send to parents. 2. Existing Tech Stack Frontend: React 19, TanStack Start v1, TanStack Router, Tailwind CSS v4, shadcn/ui. Backend & Database: Lovable Cloud (PostgreSQL / Supabase with Row Level Security, Storage Buckets). Server Execution: TanStack Start server functions (createServerFn) running on modern serverless/edge environments. 3. Core Requirements & Safeguarding Rules A. Media Ingestion & Safeguarding (Non-Negotiable) Multi-file upload: Mobile-friendly photo and video upload queue with client-side compression/thumbnail generation. Photo Consent / Safeguarding Gate: Each student in our database has a photo waiver status (forms / photo release consent). The system must allow leaders to tag students or flag any student without signed photo release so their faces are never included in parent-facing reels. B. Highlight Generation Pipeline You may propose the most reliable, cost-effective rendering engine (e.g., Remotion, Shotstack, Cloudflare / AWS Lambda with FFmpeg, or Cloudinary Video Generation): Input: 5–20 approved photos/clips uploaded by leaders that day + trip metadata (date, location, miles traveled, weather/checkpoint highlights). AI Captioning (Optional/Phase 2): Quick AI summary generated via OpenAI/Gemini Vision to create 1-2 sentence contextual overlays (e.g., "Day 2: Serving at the community garden in Denver"). Video Assembly: 30–60 seconds, vertical 9:16 format (optimised for mobile screens). Smooth Ken Burns pan/zoom effects on photos, clean video cut transitions. Royalty-free acoustic/uplifting background music bed. GoPilgrim branded intro/outro cards (Church name + Trip name). C. Leader Review & Edit Interface Leaders see a preview draft before anything is sent. Controls to: Re-order, exclude, or swap specific clips/photos. Edit or delete text overlay captions. One-tap "Approve & Broadcast to Families". D. Parent Viewing Experience Responsive mobile video player on /story and PilgrimHome parent views. Secure, tokenised sharing link so parents can download or share with family members. 4. Deliverables & Milestones Milestone Scope Deliverables Milestone 1: Architecture & Media Upload Database schema for media assets, photo release checks, and Supabase Storage upload flow. Working upload component with progress bar, compression, and student release validation. Milestone 2: Video Generation Pipeline Automated template rendering engine (Remotion/Shotstack/serverless FFmpeg). Server function generating a rendered 9:16 MP4 video from test inputs. Milestone 3: Leader Review & Approval UI Draft preview screen in GoPilgrim app with swap/reorder/caption controls. Functional leader review screen with one-tap publish state. Milestone 4: Parent Player & Delivery Parent view, responsive video player, and notification trigger. End-to-end flow tested from leader upload to parent playback. 5. Candidate Requirements Strong experience with React, TypeScript, and Supabase / PostgreSQL. Proven track record with automated video generation (Remotion, Shotstack, FFmpeg, Creatomate, or similar programmatic video platforms). Experience handling mobile media uploads, storage optimization, and background job queues. Clean code architecture that adheres to existing TanStack Start server functions and Supabase RLS. 6. Upwork Screening Questions (Add to Job Post) Which programmatic video generation tool (e.g., Remotion, Shotstack, Creatomate, AWS Lambda + FFmpeg) would you recommend for rendering 30–60s 9:16 highlight reels at scale, and why? Can you share a link or repository of a dynamic video generator or media pipeline you have built previously? How would you approach client-side image/video compression before uploading to cloud storage to keep mobile data usage low for leaders in the field?