2 months ago
London, United KingdomStaff+
Responsibilities
- Define broadly adopted standards for responsible AI, model governance, production MLOps, and agent evaluation.
- Build or validate the highest-risk portions of strategic AI systems and establish reusable standards for other teams.
- Architect AI systems that prevent systemic failure and lead resolution of severe model-safety or data-integrity incidents.
- Build reusable production-ready agent implementations, fine-tuned models, and evaluation methodologies.
- Advise on AI developments worth adopting and help set the technical roadmap.
- Serve as the senior technical partner to client leadership on AI strategy.
- Coach Senior Applied AI Engineers, delegate technical-domain ownership, and create practices that enable multiple teams to deliver safer AI.
- Contribute to recruiting and represent Scale’s AI work externally.
Requirements
- 7+ years of engineering experience with a multi-year track record owning AI/ML systems in production.
- Experience assessing training-data quality, selecting model adaptation methods, evaluating fine-tuning results, and balancing serving cost, latency, and quality.
- Experience owning an AI-powered product end to end, including direct responsibility for AI behavior.
- A track record of establishing durable evaluation methodologies, MLOps practices, or architectural patterns.
- Ability to defend technical positions with executive-level clients and leadership under pressure.
- Deep experience with regulated, sovereign, or on-premise AI deployment, including hallucination mitigation and auditability.
- Some UK assignments require BPSS screening and may require SC or DV clearance, depending on the account and candidate circumstances.
Categories
About Scale AI
Scale AI builds data annotation services and AI development tools for enterprises and government agencies, sold as a platform and managed services. Its products include the Scale Generative AI Platform for building and evaluating agents and the Data Engine for collecting, curating, and labeling training data, including RLHF and model evaluation. Founded in 2016 and headquartered in San Francisco, the company is privately held and works across domains from computer vision to LLM applications.