SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
nyra health is building software for neurorehabilitation clinics and therapists. nyra labs, the research arm, develops open models, datasets, and benchmarks for clinical speech applications. As a Software Engineer on the ML Platform team, you will design and build the infrastructure that powers the research operation—the systems that enable a small research team to run ambitious experiments quickly, reproducibly, and reliably.
You will own multiple platform domains: data pipelines for ingesting, versioning, and accessing large clinical speech datasets; distributed training infrastructure with orchestration, checkpointing, and resource scheduling; experiment management systems for configuration, tracking, and reproducibility; evaluation platforms for benchmarking and model comparison; inference optimization for cloud and on-device deployment; and release automation for packaging, validation, and open-source publishing. You'll also build internal developer tools and establish observability and access controls appropriate for sensitive clinical data.
This is not a conventional backend role. You will work directly with researchers, understand their workflows, and translate recurring bottlenecks into durable platform capabilities. The role requires strong Python skills and production-systems experience, familiarity with PyTorch, GPU workloads, and the ML development lifecycle, and hands-on experience with cloud infrastructure, containers, orchestration, and distributed computing. You should have built reliable data pipelines and worked with large, versioned datasets. Beyond technical skills, you need an operational mindset (observability, debuggability, failure recovery), platform thinking (building reusable capabilities), and research empathy (understanding that experimental workflows change quickly).
You'll be pragmatic about what needs a platform versus a script, leverage-oriented in identifying improvements that accelerate the entire team, reliable in treating reproducibility and data integrity as core requirements, self-directed in owning solutions end-to-end, and collaborative in translating experimental needs into systems. The role offers high ownership, visible impact on every experiment and release, direct collaboration with founders, and a hybrid working model in Vienna's First District.
