Pear VC

Pear VC

Member of Technical Staff, Backend - NomadicML

San Francisco · Staff+

Sponsorship not specifiedDetected 266 days ago
TypeScriptPythonGoReactNext.jsDistributed SystemsSnowflakeAWSGCPAzureKubernetesTemporalRESTgRPCMachine LearningAirflowAI OrchestrationSystems EngineeringRoboticsResearch

About the role

  • This role blends systems engineering, distributed compute orchestration, and developer experience.
  • You'll be working across cloud storage, inference scheduling, GPU clusters, and the NomadicML SDK.
  • GPU Inference Workflows: Architect pipelines to run massive multi-GPU inference jobs on foundation-scale video models, optimizing for throughput, cost, and reliability.

Responsibilities

  • Design event-driven, autoscaling job systems using Kubernetes, Pub/Sub, or Ray for analyzing terabytes of video data in parallel.
  • You'll collaborate with ML researchers to productionize their models, automate deployment and scaling, and expose those capabilities through clean APIs and SDKs used by enterprises worldwide.
  • Distributed Processing Pipelines: Design event-driven, autoscaling job systems using Kubernetes, Pub/Sub, or Ray for analyzing terabytes of video data in parallel.

Requirements

  • Deep proficiency in Python, Go, or TypeScript for backend systems.
  • Experience with AWS, GCP, or Azure (IAM, S3/Blob Storage, Batch/Compute APIs, etc.).
  • Strong understanding of GPU inference scaling, Kubernetes, container orchestration, and event-driven pipelines.
  • Familiarity with asynchronous job orchestration (Ray, Airflow, Dagster, Temporal).

Nice to have

  • Experience contributing to inference orchestration frameworks or ML infra tools (e.g., DeepSpeed, Triton, Ray Serve).
  • Understanding of video encoding, chunking, and streaming formats for efficient multi-modal ingestion.
  • Basic front-end experience (React / Next.js) for integrating backend pipelines into product workflows.
  • Background in ML infrastructure, observability, or data management systems.

Benefits

  • Build logging, tracing, and metrics pipelines that surface GPU utilization, job latency, and per-video inference health.

Company info

  • Power the NomadicML Python SDK used for programmatic video ingestion, analysis, and search - the core tool researchers and customers rely on.
  • Developer SDKs and APIs: Power the NomadicML Python SDK used for programmatic video ingestion, analysis, and search - the core tool researchers and customers rely on.

This listing is sourced directly from Pear VC's careers page and normalized into a canonical job model.