ONLINE INFRASTRUCTURE

What We Do

We enable Epic’s online services teams to build, deploy, and manage services that are used by more than half a billion players around the world. Our mission is to provide world class tools and platforms to improve the experience of our developers and make it easier, faster, and safer to build, operate, and scale their applications. We operate at massive scale as one of the largest cloud computing users in the world.

What You'll Do

Epic Games is the creator of Fortnite, Unreal Engine, and the Epic Games Store — a company that has shaped the way billions of people play, create, and connect. We build the foundational technology that powers some of the largest interactive experiences on the planet, and we give those tools to creators and developers around the world through open ecosystems.

Behind the products is an engineering organization that operates at rare scale: real-time infrastructure for hundreds of millions of players, a game engine used across film, architecture, and automotive industries, and a platform ecosystem that lets indie developers ship games to every major device on Earth.

Our AI Platform team is building the next layer of that infrastructure — an enterprise-grade stack of agentic AI systems that automates engineering workflows, accelerates developer productivity, and enables new kinds of collaboration across Epic's teams. We're not a research group and we're not deploying off-the-shelf tools. We're architecting and building production systems from the ground up, across six interconnected platforms:

• Geppetto — multi-tenant platform for team AI agents that live and collaborate in Slack channels
• EMA (Epic Managed Agents) — compute and workspace infrastructure for headless agent harness runs at scale
• Hodor — MCP OAuth gateway, plugin runtime, and governance layer for AI tool orchestration (1,500+ MAU, 700K+ tool executions)
• Multipass — agent identity, credential vault, and authorization for non-human workloads
• Vektor — org-wide memory plane with knowledge graph, deductive reasoning, and hierarchical summarization
• Roost — cryptographically signed software distribution and the Claude Code plugin marketplace

This is foundational work that will define how AI is used inside Epic for the next decade. The scale is real, the problems are hard, and the team is small enough that every engineer makes a decisive architectural impact.

As a Principal Engineer on the AI Platform team, you'll own the technical direction of our agent infrastructure stack end to end. You'll set the architecture across the six platforms above, drive alignment between them, and personally solve the hardest distributed systems and security problems that emerge as the stack scales. You'll work across teams to ensure agent identity, tool governance, memory, and execution infrastructure are coherent, secure, and operable — and you'll mentor the engineers who build alongside you. This isn't a coordinator role. You'll write production code, design protocols, make the calls that determine how agents authenticate and what they're allowed to do, and be accountable for the reliability of systems that are actively used by Epic's engineering organization.

In this role, you will

• Platform Architecture & Technical Leadership:
  • Own the end-to-end technical architecture across Geppetto, EMA, Hodor, Multipass, Vektor, and Roost — ensuring each platform is coherent with the others and that the integration seams are well-defined
  • Drive architectural decisions for agent identity and workload authorization (SPIFFE/SPIRE, OIDC, token exchange, policy planes), translating security requirements into implementable designs
  • Establish the patterns for how AI agents authenticate, receive credentials, execute tools, and are audited — and hold the bar for correctness across the stack
  • Lead design reviews for new capabilities, evaluate build vs. buy decisions, and surface technical risk before it becomes production risk
• Distributed Systems & Infrastructure:
  • Design and implement the Cluster API and provider abstractions for EMA — the layer that orchestrators depend on to launch, drive, and recover headless agent runs across Kubernetes, EC2, and other compute backends
  • Evolve Hodor's plugin runtime (WASM, gRPC sidecar, subprocess multiplexer) and its gateway security posture as external tool surface area grows
  • Architect Vektor's knowledge graph, vector search, and memory consolidation pipeline for org-wide scale — across teams, scopes, and retention horizons
  • Define durability, consistency, and isolation requirements across event-driven architectures (NATS JetStream, Redis) shared by multiple agent platforms 
• Security & Trust:
  • Lead the Multipass proposal from strategy into staffed execution — defining the separation of Hodor (human/tool) and Multipass (agent identity) and migrating the existing credential vault
  • Hold the standard for credential security across the stack: AES-256-GCM vault, AAD binding, scope isolation, default-deny policy, and audit completeness
  • Work with Epic's security organization to ensure agent-to-service trust models meet enterprise standards 
• Cross-Team Influence & Mentorship:
  • Partner with product, ML, and enterprise platform teams to shape how agent capabilities are exposed to Epic's broader engineering organization
  • Mentor senior and staff engineers across the team; conduct technical interviews and raise the hiring bar
  • Write design documents that become the reference architecture for future work, not just approvals for current work 

What we're looking for

• 12+ years of software engineering experience, with at least 4 years at staff or principal scope
• Deep expertise in distributed systems: event-driven architectures, durable execution, service mesh, and multi-tenant platform design
• Production experience with authentication and authorization infrastructure — OAuth 2.0, OIDC, SPIFFE/SPIRE or equivalent workload identity, token exchange (RFC 8693), and policy engines (OPA, OpenFGA, or comparable)
• Strong security engineering fundamentals: credential vaulting, secrets management (OpenBao/Vault), audit trail design, and least-privilege access at scale
• Fluency in at least one compiled, systems-capable language (Go preferred, Rust or C++ acceptable); comfort reading and writing Go microservices is essential given the stack
• Track record of owning multi-service platform architecture across a full product lifecycle — from design through sustained production operation
• Exceptional written communication: design documents and architecture reviews that are clear, precise, and influence without authority
• Hands-on experience building LLM-integrated systems: agent orchestration, tool-use frameworks, MCP (Model Context Protocol), or equivalent agent-to-tool middleware
• Experience with plugin or extension runtime design — WASM sandboxing, gRPC sidecar patterns, subprocess isolation, or comparable capability security models
• Familiarity with knowledge graph systems (Neo4j or comparable), vector databases, and hybrid retrieval (semantic + keyword + graph), as well as experience operating Kubernetes-based platforms: scheduling, workload identity, sidecar injection, and multi-tenancy isolation

EPIC JOB + EPIC BENEFITS = EPIC LIFE

Our intent is to cover all things that are medically necessary and improve the quality of life. We pay 100% of the premiums for both you and your dependents. Our coverage includes Medical, Dental, a Vision HRA, Long Term Disability, Life Insurance & a 401k with competitive match. We also offer a robust mental well-being program through Modern Health, which provides free therapy and coaching for employees & dependents. Throughout the year we celebrate our employees with events and company-wide paid breaks. We offer unlimited PTO and sick time and recognize individuals for 7 years of employment with a paid sabbatical.