9 months ago
New York, NY, USAMid Level / Senior
Responsibilities
- Build production AI agents for Gordian Search, Research, and Compose that produce truthful, complete, and auditable outputs.
- Design and operate evaluation infrastructure and benchmark suites that identify quality regressions before customers do.
- Own fine-tuning pipelines, including synthetic data generation, ablation studies, and domain adaptation for nuclear regulatory tasks.
- Develop frontier capabilities such as AI-accelerated weather simulations, physics-informed safety analyses, world models, vision and physics models, and agentic workflows.
- Scale and optimize model inference while debugging evaluation failures and improving latency or accuracy.
- Set technical direction for model selection, prompt design, guardrails, physics-AI integration, and the ML operations stack.
- Collaborate with NVIDIA and Microsoft researchers on tools and applications for critical infrastructure.
- Mentor and eventually hire and guide additional AI engineers.
- Deliver major production model improvements, continuous evaluation infrastructure, and at least one customer-valued frontier capability within the first 90 days.
Requirements
- 3–8 years of experience building production ML or LLM systems, including RAG, fine-tuning, evaluations, and agent orchestration.
- Demonstrated experience shipping models that users depend on in production.
- Strong command of Hugging Face, LangChain, vector databases, prompt engineering, and modern LLM operations.
- Experience designing benchmark suites, tracking model quality over time, and measuring meaningful evaluation metrics.
- Evidence of owning outcomes, setting technical direction, mentoring teammates, or leading cross-functional projects.
- Experience with physics-informed neural networks, scientific computing, or simulation acceleration is a plus.
- Published ML or AI research and contributions to open-source ML frameworks are valued.
- Deep familiarity with NVIDIA NeMo, Modulus, and CUDA optimization is a plus.
- A background in physics, engineering, or computational science is helpful.
- Nuclear industry experience is not required.
Benefits
- Top-of-market base salary, meaningful equity, health, dental, vision, FSA, and wellness stipend are offered.
- The role is fully on-site in New York City, five days per week, in Midtown/Bryant Park.
- Occasional travel to client sites, Microsoft and NVIDIA offices, or ML conferences is expected.
- The company provides a large AI research budget, compute allocation, conferences, and access to NVIDIA H100 and GH200 hardware.
- Employees collaborate with Microsoft and NVIDIA research teams and receive early access to related tools.
Categories
About Everstar
Everstar builds AI software for nuclear energy developers and operators to streamline licensing, compliance, and deployment. Its products apply language models and domain-trained reasoning to turn technical and regulatory documentation into actionable intelligence, and extend into sensor integration, computer vision, and robotics-assisted manufacturing. Founded in 2024 and headquartered in New York City, the company is privately held.