Senior Platform & Backend Engineer — AI & Data Products

About Natural Intelligence

Natural Intelligence builds AI products that turn complex, evolving data into useful knowledge and decisions. Our work spans Aura’s multimodal estate intelligence and complex claims workflows, bringing together structured records, documents, photos, video, audio and live sensor data.

We are looking for a hands-on engineer to own the secure, reliable platform and backend that make these products work. Your priorities are platform engineering first, backend product development second, and data infrastructure third. You will work directly with our founder and data lead/product owner to turn product priorities into production systems.

This is a builder role in a lean startup. You will make architectural decisions, write and deploy code, operate what you ship, and proactively resolve problems. AI-native development is fundamental to the role—not an optional productivity tool.

What you will ownGCP platform and production reliability

• Own production infrastructure on Google Cloud Platform: infrastructure-as-code, environment configuration, networking, access controls and cost management.
• Build automated CI/CD pipelines with meaningful tests, controlled releases, rollback paths and reproducible deployments. Choose managed services where they reduce operational burden.
• Establish actionable logs, metrics, traces, health checks and alerts. Define service objectives, test backups and recovery, and remove recurring sources of failure.
• Own incident response and participate in after-hours production on-call. Maintain runbooks, communicate incidents clearly and turn post-incident findings into lasting improvements.

Backend services and AI product delivery

• Design and implement maintainable backend services, business logic, integrations and documented APIs for frontend experiences and AI workflows.
• Build reliable asynchronous and event-driven workflows, including queues, retries, idempotency, timeouts, rate limiting and failure recovery.
• Productionize AI capabilities with the data lead: model and agent integrations, retrieval services, tool interfaces and background processing, with evaluation hooks, tracing and cost controls.
• Support secure multi-tenant products with robust authentication, authorization and data-access boundaries. Define API contracts and acceptance criteria with the frontend engineer and data lead/PO.

Streaming and data infrastructure

• Build dependable batch and streaming ingestion for APIs, documents, media, sensor feeds and operational systems. Handle duplicates, late events, schema changes, backpressure and replay.
• Implement storage and lake/lakehouse foundations, orchestration, data contracts, lineage and ingestion-quality checks. Support large media uploads and processing without blocking interactive services.
• Own the reliability of ingestion, storage and orchestration. Partner with the data scientist, who owns analytical transformations, feature engineering, models and insight discovery.

Security by design

• Implement least-privilege IAM, secure service identities, tenant isolation, encryption, secrets management and auditable access to sensitive data.
• Apply secure coding, threat modelling, dependency scanning, vulnerability remediation and tested authorization controls throughout delivery—not only before launch.
• Protect sensitive claims documents and other customer information through appropriate retention, deletion, logging and access policies. Build practical compliance readiness and maintain evidence of security controls.
• Secure AI integrations and agent tools: constrain permissions, validate inputs and outputs, treat retrieved content as untrusted, and prevent sensitive data exposure through prompts, logs or development tools.

AI-native engineering is a requirement

You must already have experience building AI products and developing backends through agentic coding workflows. Using a chatbot occasionally or relying only on autocomplete is not enough.

• Use coding agents such as Claude Code, Codex, Cursor or equivalent tools routinely for implementation, testing, debugging, refactoring, infrastructure changes and documentation.
• Break work into bounded tasks, provide repository context and clear acceptance criteria, and build repeatable workflows that make the whole team faster.
• Review generated code, verify behaviour with automated tests and inspect security implications. Apply appropriate approval boundaries to production changes and…