Platform Engineer (AI Agents) | IDR 30-45m p/m

Kulu; Denpasar; On-site; Confidential; Full time

Job summary

Kulu is building an AI agent platform in Bali and seeks an engineer to own the live-meeting infrastructure and reliability. You will work closely with the CTO and AI Product Engineer to ensure fast, observable, and deployable systems, tackling issues like race conditions, streaming reliability, and secure data handling.

The role focuses on maintaining the realtime pipeline, streaming sessions, and deployment tooling, with opportunities to extend into React surfaces and security-focused

Qualifications

• 5+ years of production experience in platform engineering, DevOps, SRE, or backend infra.
• Strong programming skills in Python (asyncio) and Bash scripting.
• Hands-on production experience with real-time media infrastructure (WebRTC, LiveKit, Daily, Agora).

Responsibilities

• Design, build, and maintain Kulu's real-time platform across WebRTC, LLM sessions, and Python backend.
• Own reliability, performance, and observability of core services and telemetry.
• Own the LLM streaming session layer, including connection lifecycle, streaming, and tool-call plumbing.
• Manage deployment and infrastructure across AWS and CI/CD pipelines; drive IaC (OpenTofu).
• Maintain production platform and respond to incidents; feed learnings into runbooks and IaC.
• Develop backend (Python/FastAPI/PostgreSQL/Redis) and ship frontend surfaces when needed.
• Handle sensitive meeting data with privacy and auditability.
• Contribute to architectural decisions focusing on correctness, data integrity, and security.

Skills

Python (asyncio)

FastAPI

Bash scripting

AWS (ECS)

CI/CD pipelines

Terraform/OpenTofu

PostgreSQL

Redis

Monitoring (Prometheus/Grafana)

Security best practices

React/TypeScript

English communication

Tools

LiveKit

Daily

Agora

Pipecat

WebRTC

Terraform/OpenTofu

Job description

Kulu builds an AI agents that joins live meetings — it listens, speaks, sees shared screens, and takes actions in real time. Everything users experience rides on a realtime platform: WebRTC media infrastructure, streaming multimodal LLM sessions, a Python backend, and the cloud environment it all runs in. We're hiring the engineer who owns that platform.

You'll join our Foundation team, reporting directly to the CTO, and work alongside our Lead AI Product Engineer, who owns what the agent should do — its behaviour, prompts, and evaluation. You own that the system underneath is fast, observable, correct, and deployable. When the agent misbehaves because of a race condition, a dropped audio track, a stale session, or a misconfigured environment — that's your territory, and you'll be the person who finds it.

We are deliberately moving toward a more in-person culture in Bali.

Tasks

• Design, build, and maintain the realtime platform Kulu's AI assistant runs on — WebRTC media infrastructure, streaming multimodal LLM sessions, and the Python backend behind them: the foundation every user conversation rides on.
• Own reliability, performance, and observability of Kulu's core services: the realtime pipeline end to end — rooms, agents, audio in/out, and the recording pipeline — and the telemetry that turns “the agent felt slow” into a number with a cause.
• Own the LLM streaming session layer as an engineering system — connection lifecycle, streaming, reconnection and resumption, tool-call plumbing, timeouts and watchdogs — so that behaviour changes designed by the AI product lead land on a reliable substrate.
• Own deployment and infrastructure across our AWS environment and CI/CD pipelines, and drive our infrastructure-as-code programme (OpenTofu).
• Maintain the platform in production and respond to incidents: triage, root-cause, and fix live issues across the stack, then feed every incident back into telemetry, runbooks, and infrastructure as code so it can't happen silently twice.
• Build and extend the backend (Python / FastAPI / PostgreSQL / Redis / Alembic) and ship product surfaces in our React portal when the work calls for it.
• Build services that handle sensitive meeting data with the privacy, correctness, and auditability our customers expect.
• Contribute to architectural decisions as Kulu scales, with a particular focus on system correctness, data integrity, and security.
• Bring strong engineering fundamentals to every problem: well-tested, maintainable code, clear data models, and systems that degrade gracefully under pressure.
• Maintain strong documentation and champion operational excellence across the engineering culture.
• Ship end to end: you write the migration, deploy the service, and watch the dashboards after.

Requirements

• 5+ years of production experience in a Platform Engineering, DevOps, SRE, or backend infrastructure role.
• Strong programming skills in Python (asyncio, FastAPI or similar) and solid shell scripting (Bash).
• Hands-on production experience with realtime media infrastructure: LiveKit, Daily, Agora, Pipecat, or raw WebRTC.
• Solid AWS operational experience — ECS is a must — including IAM, secrets management, VPC and networking.
• Hands-on experience building CI/CD pipelines (GitHub Actions, GitLab CI, or similar); proficiency with Git and branching strategies.
• Production experience writing and owning infrastructure as code (Terraform/OpenTofu).
• Strong grasp of distributed systems, event-driven architecture, and fault-tolerant service design.
• Solid relational database skills (PostgreSQL, schema migrations); Redis in production.
• Knowledge of monitoring and logging tooling (Prometheus/Grafana, Datadog, ELK, or equivalent).
• Strong understanding of infrastructure and application security best practices.
• Comfort across the stack: able to ship a React feature when the work calls for it.
• Strong AI-tool skills: AI is part of how you read, write, and debug code.
• Strong systems-level thinking.
• Clear communicator; fluent spoken and written English — our company language is exclusively English.

Nice to Have

• AWS certifications (e.g., Solutions Architect, DevOps Engineer).
• Security compliance experience — supporting a certification programme (SOC 2 / ISO 42001), access management, or audit preparation.
• Experience integrating streaming LLM APIs (Gemini Live, OpenAI Realtime) or speech pipelines (STT/TTS).
• Experience supporting eval or model-quality infrastructure.

Our Stack

Python (FastAPI, asyncio) · WebRTC media infrastructure · streaming multimodal LLM APIs · PostgreSQL · Redis · React + TypeScript · AWS · GitHub Actions · OpenTofu · Datadog

Benefits

IDR 35–45m/month gross

What Success Looks Like

• First month: you know our backend APIs well enough to deliver features independently, you can support all of our AWS infrastructure work, and you can debug the realtime stack — media rooms, sessions, the recording pipeline — without hand-holding.
• Three months: you've moved from supporting the platform to owning it — realtime and deployment issues route to you by default, you've shipped a measurable reliability or latency improvement, and the infrastructure-as-code programme is visibly advancing.
• Six months: the platform's configuration lives fully in code, incidents get root-caused faster because the telemetry you built points at the answer, and the AI product lead can change agent behaviour without worrying about the ground shifting underneath.

Our Operating Principles:

• Always Be Hustlin' – Move fast, outwork the competition, stay scrappy.
• Relentlessly Curious – Reason from first principles, question assumptions and explore better ways.
• Super Pumped – Show relentless enthusiasm and drive.
• Make Magic – Create experiences that delight customers.
• Obsess Over the Details – Perfect the product experience, no matter how small.
• Big Bold Bets – Think big, take risks and go after huge, transformative opportunities.
• Ownership is Key – Own the outcome, not just the task.

As Kulu grows, this role naturally expands with the Foundation team — a full-stack engineer will join on Simu, and there is a clear path to owning our infrastructure and reliability function outright.