Platform Engineer - AI Agent Infrastructure

How to apply Email hiring@sunnystep.com with subject: [Position]-[Your Full Name]. Include your CV, GitHub/portfolio, and 2-3 examples of production systems you built or shipped. Applications that do not follow this format may not be reviewed. About the role We are building Maxify, an AI-native operating platform where agents execute real business workflows through company systems while remaining accountable to source data, permissions, evaluations and measurable outcomes. We are looking for a hands-on Platform Engineer to build the reusable, multi-tenant infrastructure that makes production AI agents secure, reliable and economical to operate. You will own the shared platform beneath business-specific agents while still shipping practical capabilities end-to-end. This is not a pure infrastructure, prompt-engineering or research role. You will work across agent runtimes, backend services, data, integrations, deployment and developer tooling. What you will own Own the multi-tenant agent runtime, orchestration, state and governed memory. Build secure tool execution, authentication, authorization and customer isolation. Design approvals, audit trails, privacy controls and secrets management. Create reusable connectors for APIs, databases and business systems. Build evaluation infrastructure, testing environments and safe release gates. Implement observability for quality, failures, latency, reliability and cost. Establish deployment, versioning, backup, rollback and incident-response mechanisms. Build platform APIs, operator interfaces and developer tooling. Own cloud architecture, infrastructure automation, capacity, availability and disaster recovery. Turn repeated workflow patterns into legible, reusable platform capabilities. What we are looking for Production experience building and operating multi-tenant backend or platform systems. Strong backend and distributed-systems engineering judgment. Production proficiency in Python and/or TypeScript. Strong PostgreSQL and cloud-hosted data-infrastructure experience. Experience with APIs, queues, asynchronous jobs and event-driven systems. Strong understanding of authentication, authorization, secrets management and data isolation. Experience with CI/CD, infrastructure automation, monitoring and incident response. Experience operating production workloads on AWS, GCP or Azure. Practical knowledge of containers, networking, backups, rollback and disaster recovery. Practical LLM experience including tool use, structured outputs, context management and evaluation. Ability to debug across application, infrastructure and external-service layers. Clear written communication and sound architectural judgment. Strong signals You have shipped and operated a production agentic or automation platform, not only a demo. You have built multi-tenant systems with explicit isolation and authorization controls. You have designed evaluation, observability, approval or audit systems for AI workflows. You have owned reliability and cloud infrastructure while remaining product-oriented. You use AI development tools extensively while validating outputs and retaining technical ownership. You can explain a system you built end-to-end, what failed and how you prevented recurrence. Nice to have Experience with OpenAI Agents SDK, LangGraph, OpenClaw or comparable systems. Experience with Supabase or PostgreSQL at production scale. Experience with Lark/Feishu, Shopify, CRM, accounting or enterprise SaaS integrations. Familiarity with knowledge graphs, ontology design or governed memory. Early-stage startup or founding-engineer experience. Current stack OpenClaw, Python, TypeScript, Supabase/PostgreSQL, Lark/Feishu APIs and CLI, Shopify and other business-system APIs, Git-based development, automated testing, deployment controls, monitoring and rollback. What this role is not Prompt writing without software ownership. Pure machine-learning research or model training. Pure DevOps, Kubernetes or cloud administration. Frontend-only or backend-only feature delivery. Building abstractions for hypothetical scale before real workflows require them. Success in the first 30 days Complete an architecture, security and reliability review and identify the major risks and delivery bottlenecks. Operate the current agent runtime and deployment environment independently. Ship at least one meaningful production platform improvement. Deliver one reusable platform capability with tests, telemetry and rollback. Establi…