⏰ Start date: ASAP
📍Zürich, Switzerland (on-site, remote not possible)
🦾 Full-time (100%)
Your roleAs an AI infrastructure SWE you will build the systems that underpin our robot learning. You will work across data pipelines, internal tooling, and model deployment from day one as we build the foundations of our ML infrastructure.
What you’ll be doing:
Build a tiered data processing platform from raw ingestion to versioned training dataset generation
Build and operate the training infrastructure by using containerized deployments and cloud GPU provisioning
Ship models to production with cloud and edge inference and build the evaluation harness to guarantee safe deployments
Create and maintain internal data quality and inspection tooling
What you should have:
5+ years of experience in a professional SWE environment building production software with a significant focus on data platforms or ML infrastructure
Strong Python knowledge and comfortable in a typed language (Rust, Go, C++, ...)
Experience in data pipelines and storage: tiered architecture, workflow orchestration, backfills, and schema evolution
Cloud training experience: you have provisioned GPU instances and trained in a reproducible setup, from containerized deployments to a model registry
Hands-on ML experience: you have trained models and understand dataloader throughput, GPU utilization, and can debug slow or stalled training runs
Strong SWE foundations: You work with IaC and code reviews, propose architectural changes and refactors, and build internal tooling and automation
These skills are a plus:
Edge inference deployment (Jetson or similar) with TensorRT, ONNX, quantization
Multimodal and time-series data: video pipelines, sensor logs, MCAP, time alignment across sources
Distributed training and training performance optimization
GPU cluster management and job orchestration
Rust in production
Don't worry if you don't hit every check-mark. We value people who learn fast and care about building great products. Just give it a go and apply.
Skills Required
- 5+ years professional SWE experience building production software with focus on data platforms or ML infrastructure
- Strong Python knowledge
- Comfortable in a typed language such as Rust, Go, or C++
- Experience in data pipelines and storage: tiered architecture, workflow orchestration, backfills, and schema evolution
- Cloud training experience: provisioned GPU instances and trained in reproducible setup, from containerized deployments to a model registry
- Hands-on ML experience: trained models and understand dataloader throughput, GPU utilization, and debug slow or stalled training runs
- Strong software engineering foundations: Infrastructure as Code, code reviews, proposing architectural changes, building internal tooling and automation
- Edge inference deployment (Jetson, TensorRT, ONNX, quantization)
- Experience with multimodal and time-series data, video pipelines, sensor logs, MCAP, and time alignment across sources
- Distributed training and training performance optimization
- GPU cluster management and job orchestration
- Rust in production
What We Do
Loki Robotics develops autonomous robots designed to free people from repetitive chores and support facility operations. Its first robot cleans bathrooms in commercial facilities, while the company builds autonomy infrastructure for places where people live, work, and gather. Founded by ETH Zurich alumni, Loki aims to deploy intelligent machines that handle routine work, beginning with commercial cleaning and expanding toward broader autonomous facility management.
