ML Infrastructure Engineer — Scale Training & Serving (Hybrid)

Acceler8 Talent; Cambridge (MA); Hybrid; USD 140,000 - 210,000; Full time

Job summary

Acceler8 Talent partners with an early‑stage AI company in Cambridge, MA, building infrastructure for advanced AI research and discovery. The role bridges research and production, focusing on scalable ML training and inference systems in a hybrid setting.

As a Member of Technical Staff – ML Engineer, you design distributed training, GPU scheduling and model‑serving services, turning research code into robust production tools for researchers to train, deploy and evaluate models at scale in

Qualifications

• 3+ years of experience building and operating production ML infrastructure.
• Experience in distributed systems, ML platforms, model serving or training infrastructure.
• Experience developing tools used by researchers or engineers.
• Ability to diagnose and resolve complex production issues.
• Familiarity with GPU computing, cloud infrastructure or ML performance optimization is beneficial.

Responsibilities

• Design and build scalable ML training and inference infrastructure
• Develop distributed training systems, GPU scheduling and model-serving services
• Turn research code into reliable, production-ready services and tools
• Build infrastructure supporting AI agents, model evaluation and experimentation
• Troubleshoot complex infrastructure failures and improve system performance and reliability

Skills

Distributed systems

ML platforms

Model serving

Training infrastructure

Job description

Acceler8 Talent partners with an early‑stage AI company in Cambridge, MA, building infrastructure for advanced AI research and discovery. The role bridges research and production, focusing on scalable ML training and inference systems in a hybrid setting.

As a Member of Technical Staff – ML Engineer, you design distributed training, GPU scheduling and model‑serving services, turning research code into robust production tools for researchers to train, deploy and evaluate models at scale in