Senior/Staff AI Scientist
Tonic AI · Remote · Remote ·
COMPENSATION
€•• – ••k
Listed on the employer's job board (Ashby)
- Senior/Staff level
- Fully remote
- LLMs & Agents
- Synthetic Data
- Production ML
You will design systems that generate coherent synthetic environments for agent training and build synthesis models for large-scale data de-identification. The work ensures real enterprise data remains useful and safe for AI model training while maintaining statistical fidelity.
Your responsibilities include improving entity detection accuracy, building evaluation infrastructure for agent outcomes, and fine-tuning open-weight models. You will partner with leading AI labs to solve complex data problems and optimize model inference for sensitive customer environments.
What you'll do
- Design systems for generating longitudinally coherent synthetic environments for agent training
- Build synthesis models that preserve format, distribution, and semantic consistency at scale
- Train and improve NER models for entity detection across free text and structured fields
- Create evaluation infrastructure that grades agent outcomes on real tasks
- Fine-tune open-weight models on generated data to guide product direction
- Optimize model inference for efficiency on large volumes of sensitive data
- Expand coverage into new domains, languages, and entity types for diverse data formats
- Partner with frontier labs and enterprise teams to ship model improvements
What you'll need
- 5+ years building production ML systems or PhD with 1+ year experience
- Deep expertise in LLMs, agents, RL, NER, or information extraction
- Proven track record of training and shipping models to production
- Strong software engineering fundamentals and ability to write maintainable code
- Fluency with PyTorch, distributed training, and standard agent frameworks
- Experience handling messy, sensitive real-world data with privacy constraints
- Ability to frame ambiguous problems and drive them to measurable results
- Experience with generative models where output fidelity and utility are critical
Nice to have
- Background in synthetic data generation
- Experience with data privacy or de-identification techniques
- Expertise in benchmark construction for AI models
Stack
- PyTorch
- LLMs
- Agents
- NER
- Distributed Training
How you'll work
- Remote-friendly work environment
What you get
- Competitive salary and equity package
- Unlimited paid time off
- 401k plan with employer contribution
- Medical, dental, and vision insurance
- Generous parental leave policy