xelys jobs xelys jobs

Machine Learning Infrastructure Engineer

WindBorne Systems

on-sitemidpermanentbackenddevops Redwood City, CA 128 days ago via LinkedIn
140,000 - 240,000 USD/annual

See how well this job matches your profile

Sign up to get an AI match score and generate a tailored application in seconds.

Get your match score

Tags

Machine LearningPyTorchDockerDistributed TrainingData PipelinesLoggingMonitoringCloud Cost OptimizationGPU ClustersGeospatial/Weather Data

About the role

Role Overview

Machine Learning Infrastructure Engineer at WindBorne Systems. You will reduce operational load on a fast-moving research team by building reliable end-to-end infrastructure that powers real-time AI weather models.

Responsibilities

  • Research-to-Operations pipelines: Own uptime end-to-end for real-time forecasts with strict latency requirements.
    • Build health monitoring and improve logging.
    • Diagnose failures across distributed nodes.
  • Inference scaling & compute strategy:
    • Evaluate cost/performance tradeoffs across cloud providers as production deployment scales.
    • Help manage and grow on-prem compute and storage resources.
  • Data pipelines & upstream reliability:
    • Build training and real-time data pipelines across many upstream sources (satellites, government agencies, and balloon observations).
    • Handle upstream delays, incomplete/variable documentation, and changing data quality.
    • Implement data QC checks plus logging and alerting for edge cases.
  • Training infrastructure:
    • Make distributed training reliable (e.g., silent OOMs, network faults, storage issues).
    • Add monitoring, auto-recovery, and job scheduling so researchers can run experiments with less babysitting.

Requirements

  • Experience running production ML systems (beyond reactive firefighting—systems that remain stable).
  • Experience with large datasets and fast release cycles.
  • Hands-on experience with PyTorch and Docker, including debugging issues related to memory management and network saturation.
  • Familiarity with techniques such as compression (as relevant to the stack).
  • Strong systems mindset and ability to bring structure and clear processes in a research-heavy environment.

Nice to Have

  • Experience with weather data, geospatial pipelines, or scientific computing.
  • Experience with petabyte-scale datasets.
  • Experience managing GPU clusters or job schedulers.

Location & Work Model

  • In-person required in Redwood City, CA (1600 Bridge Pwky).

About WindBorne Systems

WindBorne Systems builds AI-driven weather forecasting by designing, manufacturing, and operating its own smart weather balloons. The company uses the collected atmospheric data to produce high-frequency global forecasts with strict real-time requirements, and focuses on turning research into reliable production infrastructure.

Scraped 7/18/2026