Senior/Staff Software Engineer, Agent

🎨

About OpenArt

OpenArt is an AI storytelling and visual creation platform used by millions of creators worldwide. We're building the next generation of creative tools powered by cutting-edge AI — enabling anyone to create videos, visuals, characters, and stories with unprecedented speed and imagination.

We believe the future of creativity is AI-native, and we're shaping that future.

🚀

Why Join OpenArt

• Ground floor of a new category — help define what "agent" means for creative work, the way Director redefined what "directing" means for video. You'll set the architecture everything else is built on.
• Own the whole agent layer — not a service, not a slice. Harness, orchestration, memory, evals, tooling: the entire system is yours to design and lead.
• Real scale, real stakes — your agents run for millions of creators, and every improvement in reliability, latency, or quality shows up in the product within days.
• Founder-led engineering culture — both founders are technical and deeply involved in product and architecture.
• AI-native product — you'll decide how frontier models get orchestrated, constrained, and exposed as experiences that feel effortless rather than technical.
• High ownership, low process — we value judgment, clarity, and speed over bureaucracy.

🎯

About the Role

We're hiring a senior or staff engineer to lead the agentic architecture behind OpenArt's next generation of creative products, starting with Director — which lets anyone "vibe direct" a multi-minute cinematic video by chatting with AI. You'll own the agent infrastructure that pushes experiences like this further: longer-horizon planning, tool use across multiple model vendors, memory across sessions, sub-agent orchestration, and the harness that keeps all of it reliable in production.

This isn't a research role — it's a product-and-systems leadership role. You'll define the architecture of the agent harness (tool use, context engineering, multi-step planning, orchestration, evaluation), set the engineering standards the team builds against, and stay hands-on: building the MCP servers and CLI tooling around it, and shipping the frontend experiences on top of it.

You'll work directly with the founders and design to turn "what should this agent be able to do" into a system that actually does it — reliably, at scale — and you'll be the technical voice that decides how.

🛠

What You'll Do

• Own the end-to-end architecture of the agent harness that powers OpenArt's conversational and autonomous creative workflows: tool use, context/memory management, multi-step planning, and sub-agent orchestration.
• Design how agents plan and execute long-running creative tasks — scene planning, storyboarding, asset generation, revision loops — across multiple model vendors, with clear failure modes and recovery paths.
• Build and maintain MCP servers and CLI-based agent tooling that connect our agents to internal services, external model vendors, and creative asset pipelines.
• Author reusable Agent Skills and structured workflows, and define the patterns other engineers use to add new ones.
• Build the evaluation and observability stack — automated evals, regression suites, tracing, and quality gates — so the team knows when an agent is working well and catches regressions before users do.
• Drive reliability, latency, and cost per successful run as first-class metrics, not afterthoughts.
• Ship polished, production frontend experiences (React/Next.js) for agent-driven products, end to end.
• Set the technical bar for agent engineering at OpenArt: architecture reviews, engineering standards, and mentorship as the team grows around you.
• Track the fast-moving agent ecosystem (MCP, Agent Skills, orchestration frameworks, new model capabilities) and decide what belongs in OpenArt's stack — and what doesn't.
• Partner closely with founders, product, design, and GTM to translate user needs into technical direction, and communicate trade-offs clearly.
• ⚙

What We're Looking For

• Core Requirements
• 7+ years of full-stack engineering experience shipping and owning production systems at scale, including 2+ years building LLM-powered agents in production — not just prototypes or demos.
• Deep, hands-on experience with agent architecture: tool use, context/memory management, multi-step planning, sub-agent orchestration, and long-running workflows.
• Track record of designing evaluation frameworks and observability for agentic systems — you know how to measure whether an agent is working and how to keep it from regressing.
• Strong system design and data modeling skills — comfortable reasoning about state, schemas, and APIs for real product use cases.
• Experience authoring reusable Agent Skills, structured agent workflows, or equivalent abstractions that other engineers build on.
• Demonstrated technical leadership — you've set direction for a system or team, and people follow your architectural calls because they've been right before.
• A strong AI-assisted coding practice — you use tools like Claude Code, Codex, or Cursor as a core part of how you build, and have real opinions about what makes them work well.
• High product sense — you think about what makes an agent experience feel trustworthy and delightful, not just functional.
• Strong communication skills — you can explain agent behavior and technical trade-offs clearly to founders, designers, and non-technical teammates.
• Nice to Have
• Practical experience building MCP servers and CLI tooling for agents.
• Experience with agent orchestration frameworks (Claude Agent SDK, LangGraph, or similar) and LLM evaluation / observability tooling.
• Hands-on experience with image / video generation models and creative asset pipelines.
• Experience with guardrails, safety, and cost controls for autonomous systems.
• Startup experience or ownership of a product surface from 0 to 1.
• 💰

What Success Looks Like in 3–6 Months

• The agent harness has a clear, documented architecture that the team builds against — and it's yours.
• A production agent workflow you designed is shipped and running for real users inside OpenArt's ecosystem.
• An eval and observability loop exists, and regressions get caught before users see them.
• Reliability, latency, or cost per successful run has measurably improved on a workflow that matters.
• You've formed a tight feedback loop with design, founders, and real users.
• You're setting the direction for agent engineering — pushing ideas, not waiting for specs.

If you want to lead the architecture that turns frontier models into creative agents millions of people use — and own it from day one — we'd love to talk.