Staff AI/Machine Learning Engineer

This is a fully remote position, open to applicants in United States.

📋 Description

• Create and develop systems that produce longitudinally consistent synthetic environments for agent training and assessment, which include persona modeling, task generators, and verifiable ground truth.

• Construct and sustain synthesis models that deliver realistic replacement values at a large scale, maintaining format, statistical distribution, and semantic integrity to ensure de-identified data remains useful downstream.

• Train and enhance the NER models that support our entity detection, improving accuracy and recall across free text, structured fields, and mixed enterprise data at scale.

• Develop evaluation infrastructure that assesses agent outcomes rather than just traces, enabling true differentiation between frontier models on actual tasks.

• Fine-tune and assess open-weight models using Tonic-generated data, translating benchmark results into product and research strategies.

• Broaden coverage into new domains, languages, and entity types, while managing the diverse formats and edge cases that arise from real customer data.

• Oversee model evaluation comprehensively: precision and recall for detection, utility preservation for synthesis, and outcome-level grading for agents.

• Enhance inference so models operate efficiently on large amounts of sensitive data within customer environments.

• Collaborate directly with frontier labs and the enterprise ML team to transform challenging data issues into implemented model enhancements.

• Establish the technical direction for a small, senior team, elevating standards for rigor, reproducibility, and successful delivery.

⛳️ Requirements

• 8+ years of experience (or a PhD with 3+ years) in building production ML systems, demonstrating significant expertise in areas such as LLMs, agents, RL, NER, or information extraction.

• Practical experience in training and deploying models to production, with a pragmatic understanding of quality: you can assess it, recognize its failures, and know when it’s ready for release.

• Background in generative or synthesis models where both output fidelity and downstream utility are important, rather than mere plausibility.

• Strong foundation in software engineering principles. You write code that others can build upon.

• Proficiency with contemporary training and evaluation frameworks (PyTorch, distributed training, standard agent and benchmark frameworks).

• Comfort in handling messy, sensitive, real-world data along with the associated privacy considerations.

• A proven history of articulating ambiguous challenges and steering them towards measurable, delivered outcomes.

• Bonus: experience in synthetic data generation, data privacy or de-identification, or benchmark development.

🏝️ Benefits

• Competitive salary and equity.

• Unlimited paid time off.

• 401k plan with employer contribution.

• Medical, dental, and vision insurance.

• Generous parental leave policy.

• Remote-friendly work environment.