ML Engineer, Post-Training and Evaluation (AI)

TL;DR

ML Engineer, Post-Training and Evaluation (AI): Adapting open-weight models for enterprise customers through fine-tuning and evaluation with an accent on SFT, preference optimization, and RLHF. Focus on building evaluation harnesses, creating reproducible data pipelines, and deploying adapted models to production.

Location: On-site in San Francisco or New York. Relocation support is provided.

Company

Developing open-weight superintelligence models for individuals, agents, and enterprises.

What you will do

• Fine-tune open-weight models for specific customer use cases using SFT, DPO, and RLHF.
• Design and maintain evaluation infrastructure, including eval suites and test set curation.
• Develop reproducible data pipelines to clean and format raw customer inputs.
• Debug training and inference issues by analyzing loss curves and training dynamics.
• Deploy fine-tuned models across public cloud, VPC, and on-premises environments.
• Establish best practices and benchmarks for the company's fine-tuning and evaluation playbooks.

Requirements

• 3+ years of engineering experience with significant exposure to applied ML or MLE.
• Hands-on experience with LLM fine-tuning, including dataset preparation and training loops.
• Strong software engineering fundamentals in Python.
• Proficiency with GPU compute management and training infrastructure.
• Experience working in customer-facing environments to translate requirements into training strategies.
• Must be based in or relocate to San Francisco or New York

Culture & Benefits

• Top-tier salary and equity package.
• Comprehensive medical, dental, vision, life, and disability insurance.
• Fully paid parental leave and financial support for family planning.
• Daily provided lunch and dinner.
• Relocation support and regular team off-sites.
• High-agency environment within a small, talent-dense team of researchers from DeepMind, OpenAI, and Meta.