Set the quality bar: Establish offline evaluations, simulated scenarios, safety tests, and human-calibrated LLM-as-judge evaluations integrated into development and release workflows.
Own production excellence: Design service-level objectives, fallbacks, human handoff, canary releases, and safeguards against prompt injection, jailbreaks, hallucinations, and exposure of personal information.
Lead through technical influence: Set strategy across multiple engineering teams, mentor senior engineers, and align Product, Design, and Operations on coverage, localization, and rollout decisions.
What We’re Looking For
Minimum qualifications
10+ years building production ML or AI systems, including 4+ years leading complex ML initiatives end to end.
Deep expertise in LLM-driven systems, including inference optimization, prompt and program design, model adaptation, safety controls, and evaluation.
Strong software engineering skills in Python and at least one of Go, Java, or C++, with hands-on experience in distributed services and production infrastructure.
Demonstrated ownership of highly available services, including service-level objectives, incident response, on-call leadership, and post-incident improvements.
A track record of shipping customer-facing intelligent experiences with measurable impact through A/B testing and rigorous metrics.
Ability to set technical direction across teams, lead architectural decisions, and mentor experienced engineers.
Preferred qualifications
Experience building production agentic architectures using planning, execution, memory, and multi-step reasoning.
Expertise in retrieval-augmented generation over complex, policy-heavy knowledge bases and integrating agents with structured workflows.
Experience with voice agents, including streaming ASR/TTS, barge-in, endpointing, telephony, WebRTC, and conversational quality evaluation.
Experience building support automation for large consumer platforms, including routing, policy codification, agent assistance, and automated resolution.
Experience with multilingual language understanding and generation, code-switching, or low-resource languages.
Practical experience with fine-tuning, distillation, or parameter-efficient adaptation techniques such as LoRA.
Experience with hallucination mitigation, safety red-teaming, privacy-conscious system design, and controlled production rollouts.