Founding Engineer

About Cozmo AI

We build AI Employees that speak, think, and act across revenue-critical workflows for regulated enterprises. Real calls, real outcomes, real accountability. Our Agent Protocol Engine logs every step, tool call, and reason code, so customers can trust what the agent did and why. Backed by Y Combinator, Goodwater Capital, and Dubai Future District Fund, we’ve reached $1.5 M ARR in < 6 months, and are scaling from $1.5 M → $10 M ARR in 12 months.

What you will own

• Delivery at scale: Hit 99.9 percent uptime across voice and API services. Keep median call round trip latency under 600 ms. Support 1,000+ concurrent calls with graceful degradation.
• Platform and reliability: Own core services across LiveKit, SIP ingress and egress, STT, LLM routing, TTS, memory, and orchestration. Productionize observability, on-call, incident response, SLOs, and error budgets.
• Security and compliance: Drive SOC 2, HIPAA, UAE data residency, and tenant isolation. Build audit trails and redaction at the platform level.
• Cost and performance: Reduce cost per minute and per interaction while improving quality. Set and track infra unit economics by workload and client.
• Team and culture: Hire, coach, and level up an A team across platform, voice pipeline, applied AI, data, and forward-deployed engineering. Set a high-urgency, high-ownership culture.
• Partner engineering: Work with sales and delivery to land and expand enterprise accounts. Turn bespoke wins into reusable product and infra.
• Roadmap: Translate business goals into a clear, staged plan. Sequence bets, retire tech debt, and keep the platform boring where it should be.

What you will do in your first 90 days

• Ship a Latency and Stability Plan with concrete targets by region and carrier, plus weekly scorecards.
• Stand up golden paths for new client deployments with Terraform, templates, and runbooks.
• Implement auditability by default across the Agent Protocol Engine. Every action logged and explainable.
• Launch a weekly Reliability Review. Top incidents, learnings, and fixes closed in the same week.
• Hire or uplevel 3 to 5 key ICs in platform, real-time media, and infra.
• Propose a cost model with per-call and per-minute margins, and a plan to improve it by 20 percent.

Our stack today

• Real-time: LiveKit SFU, SIP trunks, TURN, WebRTC
• Pipeline: STT, LLM, TTS with smart caching and interruption control
• Data and state: Kafka or MSK, Redis, ClickHouse, Postgres
• Orchestration: Temporal, Kubernetes on EKS, Helm, Terraform
• Observability: Prometheus, Grafana, OpenTelemetry, Langfuse
• Compliance and privacy: regional isolation, KMS, encryption in transit and at rest

What great looks like

• You have led 5+ engineers. You can still dive into a p99 trace and fix it the same day.
• You have shipped real-time systems or high-throughput infra where every 100 ms matters.
• You have owned an on-call rotation and driven SLOs, postmortems, and steady incident burn-down.
• You have turned custom enterprise asks into product building blocks.
• You care about engineering taste. Simple designs, clear interfaces, clean rollback stories.

Qualifications

• 3+ years building distributed systems or real-time products
• Deep familiarity with at least two of these: WebRTC or VoIP, media pipelines, low-latency streaming, vector and hybrid search, workflow engines, multi-region Kubernetes
• Strong command of cloud networking, load balancing, and traffic shaping
• Experience with security by design and audits in regulated settings
• Excellent hiring, performance management, and coaching skills
• Comfortable working shoulder to shoulder with engineering, customer success, and founders
• Graduated from a top Ivy school

Nice to have

• Speech systems experience across STT and TTS or voice UX
• Prior work in BFSI, healthcare, or other regulated industries
• Multi-tenant architecture with strict data residency controls
• Cost optimization at seven-figure annual infra spend

How we work

• High urgency, high ownership, and clear weekly goals
• Write decisions down, keep interfaces small, ship daily
• Respect for deep work and a strong on-call discipline
• We celebrate wins and we fix what break