HeyMilo is a platform that helps recruiters deploy multimodal AI agents to interview and evaluate candidates at scale.

This role is for an exceptional full stack engineer who wants to work on frontier AI technology including real-time communication, low-latency pipelines, orchestration, and production infrastructure that supports interviews happening across the world.

If you want a very predictable roadmap and narrow responsibilities, this role is not for you.

If you want to build real-time AI systems that feel human, keep reading.

What you'll work on

• Building real-time AI interview systems (voice + video, low latency)
• Designing and scaling infrastructure for thousands of concurrent interview sessions
• Developing backend APIs and workflow orchestration (FastAPI, Temporal)
• Shipping full stack product features used by employers and candidates every day
• Building queue-driven systems for job execution, retries, and automation (SQS)
• Working on retrieval systems and prompt-driven agent workflows (VectorDBs, MCP)
• Improving reliability, performance, and observability across production systems
• Supporting enterprise integrations, data pipelines, and security patterns

What we're looking for

• Use Cursor/Devin to write at least 50% of your code
• 3–5+ years of professional software engineering experience
• Full stack engineer who can ship end-to-end and operate systems in production
• Strong systems thinking and engineering fundamentals
• Comfortable building low-latency, real-time systems
• Basic understanding of prompt writing + ML fundamentals
• High ownership, high accountability, low ego
• Deep belief in what were building

Tech you should be comfortable with

• Backend: Python, FastAPI, async systems
• Frontend: ReactJS, NextJS, ShadCN
• Infra & deployment: AWS, GCP, Docker, Docker Compose, Kubernetes
• Queues & orchestration: SQS
• Databases & analytics: MongoDB, ClickHouse,
• Retrieval: VectorDBs, RAG-style systems
• Real-time: LiveKit, Pipecat, low-latency audio/video pipelines
• AI agent tooling: MCP servers, prompt-driven workflows

This role is not for you if

• You want a strict 9–5 job
• You prefer narrow scopes and fixed task lists
• You avoid production responsibility
• You want safety over upside