Platform Engineer in London

Terra is the connective tissue between the world's health data and the developers building on top of it. We ingest from 500+ upstream sources, normalize everything into a single schema, and deliver 8B+ events/year to 1,000+ developers and AI labs via webhooks, SQL, cloud storage, and queues. The platform is the product. We're building the upstream supplier connector — a new ingestion layer for data suppliers to push directly into Terra. AI labs are consuming at rates we didn't plan for. The pipe needs to get wider, faster, and more reliable. This role exists because the platform needs a dedicated owner. Someone who thinks end-to-end — from a Garmin syncing at 3am to a webhook landing 400ms later. You're not building features. You're building the foundation.

What you’ll own:

• The full data path from source to destination.
• Ingestion Pipeline— How provider data enters Terra from cloud APIs (Garmin, Fitbit, Oura) and mobile SDKs (Apple Health, Health Connect). Raw data in, queued for normalization.
• Normalization Engine— Transforms heterogeneous provider payloads into Terra’s standardized models: Activity, Sleep, Daily, Body, Menstruation, Nutrition. The core IP that makes 500+ sources feel like one.
• Event Delivery— Webhooks, Postgres/MySQL, Supabase, S3, SQS, Kafka. At-least-once delivery, ordering, retries, dead-letter queues. Data reaches destinations reliably.
• Provider Framework— How new Sources get onboarded. Web-based (OAuth, polling, rate limiting) and mobile-only (on-device SDK, background sync). Adding provider #501 should be as fast as #5.
• Auth & Connections— Connect widget, custom UI flows, OAuth sessions, Terra User lifecycle, Reference ID mapping.
• Upstream Supplier Connector— New ingestion surface. Data suppliers push into Terra instead of us pulling. Ground-up build.
• API Versioning— Backwards compatibility and deprecation across the /v2 surface.
• Observability— Monitoring and alerting across the full pipeline. You know when a provider degrades before our developers do.

What you’ll build:

• Scale event delivery to 10x— AI lab consumption is growing exponentially. Redesign the hot path without proportional cost increase.
• Ship the upstream supplier connector— Schema contracts, auth, validation, rate limiting. The interface for suppliers to push directly into Terra.
• Redesign normalization for schema evolution— Providers change APIs. Our models evolve. Downstream consumers can’t break.
• Multi-destination fan-out— One event, reliably delivered to webhooks, SQL, S3, and queues simultaneously. Independent retry logic per destination.
• Provider health dashboards— Real-time visibility into every Source’s freshness, latency, error rates, and schema drift.
• Harden mobile SDK data path— Apple Health and Health Connect are fundamentally different from cloud APIs. On-device processing, background sync constraints, OS-level limits. Make it scale.

Who you are:

• A systems thinker — you see data flow and failure modes before you see features.
• Battle-tested at scale — you’ve operated platforms processing billions of events. You’ve fixed production at 2am.
• Observability-first — monitoring and tracing are part of your design, not bolted on after.
• Opinionated where it matters — event delivery guarantees, schema evolution, idempotency, API versioning. Opinions earned by shipping.
• Fast — architecture whiteboard Monday, production deploy Friday.
• Default yes — hard problems make you lean in.
• Accountable — you own production, you respond to incidents, you care about uptime. Skin in the game.

Big Plus:

• You are an athlete. You train, you compete, you push limits — or at the very least, you are obsessed with quantifying your own data. The discipline, ambition, and courage it takes to show up every day and get better is the same energy we run on. If you understand the data because you live it, you’ll build a better product.