AI Systems Engineer: Scalable Data & Automation

Proximal; San Francisco, Northern (CA, KY); Hybrid; USD 140,000 - 210,000; Full time

Job summary

Proximal is building the research systems needed to identify what models can't yet do, build the tasks required to teach them, measure improvements, and continuously produce the data frontier models need. We are assembling an early team with deep experience building coding agents, RL infrastructure, and scalable systems.

As a software engineer, you'll develop the core infrastructure powering our data creation engine, focusing on reliability and automation in a research-driven setting.

Qualifications

• Experience designing and building large-scale systems with reliability and performance.
• Ability to work in research-heavy environments and handle ambiguous technical problems.
• Skill in building automations and agent-based systems.
• Strong system design judgment for tradeoffs in performance, reliability, and cost.
• Self-directed with ownership and rapid execution without heavy oversight.

Responsibilities

• Build infrastructure to run hundreds of thousands of concurrent agents reliably for 24+ hours.
• Create ephemeral, production-like multi-node training environments with snapshotting and restore capabilities.
• Develop automated QA to adversarially evaluate tasks and agent outputs for correctness, fairness, and reward hacking.
• Build systems to continuously index all code on the internet.

Skills

Systems design

Reliability at scale

Research-friendly mindset

Automation / agent systems

Experimentation

Ownership

Job description

Proximal is building the research systems needed to identify what models can't yet do, build the tasks required to teach them, measure improvements, and continuously produce the data frontier models need. We are assembling an early team with deep experience building coding agents, RL infrastructure, and scalable systems.

As a software engineer, you'll develop the core infrastructure powering our data creation engine, focusing on reliability and automation in a research-driven setting.