We launched recruiters.rip, let your AI apply for you. First month free with code onemonthfree

Founding Machine Learning Engineer

1 month agoLeadSalary: From 230000 USD / yearSan Francisco, USAOnsiteFull TimeAiJobs by Arcade

Arcade is an enterprise AI-agent actions runtime for governing, authorizing, executing, and auditing agent tool calls.

Series ARecently funded20 current maintainers20 active leads5 new active leads5 lead step-downs
Burlingame, United States
About Arcade

Arcade AI, Inc. provides runtime infrastructure that lets AI agents securely access business systems under user-specific permissions, with policy enforcement, tool execution, and centralized auditability.

View jobs by Arcade

Skills

Candidate Availability

About the Role

You will own the end-to-end machine learning pipeline, from data ingestion and training through evaluation, release, and production monitoring. You will build and fine-tune models for tool selection, routing, retrieval, and agent memory; create evaluation systems; and deploy optimized models to customer VPCs and air-gapped environments. You will turn telemetry into training data, choose the ML stack, make build-versus-buy decisions, and shape the ML strategy and future hiring.

Requirements

  • 7+ years of software engineering experience
  • 4+ years training and shipping production ML systems
  • Experience training or fine-tuning models that reached production and improved customer-relevant metrics
  • Expertise in production agent systems, including harnesses, memory, skills, tool use, and sub-agents
  • Knowledge of fine-tuning and deriving training data from raw telemetry
  • Statistics fluency for evaluating whether metric changes are meaningful
  • Experience deploying models under latency, GPU, or infrastructure constraints
  • Python for training and ML work
  • TypeScript or Go for production model-serving services
  • Ability to make and defend technical stack decisions

Responsibilities

  • Own the end-to-end ML training pipeline from data through evaluation and release
  • Train and fine-tune models for tool selection, routing, retrieval, and agent memory
  • Expand model use cases for tool and agent-context search and recommendation
  • Build offline and online evaluation systems and compare models with Claude, GPT, and Gemini
  • Quantize, optimize, package, and serve models for customer VPCs and air-gapped environments
  • Turn production agent traces and tool-call telemetry into training data within enterprise data boundaries
  • Choose the ML stack, write the strategy, and inform ML hiring
  • Use expertise in agent systems to decide where models improve agents
  • Use AI tools to increase delivery speed

Benefits

  • Equity
  • Benefits