AI Research Engineer – Agentic Post-training

This is a fully remote position, open to applicants anywhere in the world.

📋 Description

• Execute comprehensive research and engineering projects to enhance post-training of agentic and tool-utilizing models in order to attain state-of-the-art (SOTA) outcomes.

• Propel extensive, cross-disciplinary model enhancements, focusing on areas such as factual accuracy, adherence to instructions, usage of tools/functions, coordination among multiple agents, and calibration of reasoning.

• Create and refine large-scale post-training systems, which encompass data pipelines, training workflows, evaluation frameworks, and benchmarking infrastructure.

• Establish robust evaluation suites and diagnostic instruments to gauge model readiness for deployment.

• Fortify feedback mechanisms from real-world product utilization, integrating both explicit and implicit user feedback into post-training processes.

• Partner with tooling, product, and training teams to enhance the utility, dependability, and agentic functionalities of cutting-edge models.

• Work closely with research, engineering, and cross-functional teams to identify which integrations are ready for production inclusion in significant model releases.

⛳️ Requirements

• Bachelor’s degree in Computer Science, Machine Learning, or a related discipline; an advanced degree (MS/PhD) is preferred, along with a solid publication history in leading AI conferences.

• Experience with multimodal post-training processes and data pipelines, especially for agentic systems and tool usage.

• Practical experience in applying post-training techniques at scale using distributed training frameworks (e.g., multi-node GPU settings).

• Proven ability to enhance model functionalities in aspects like reasoning, tool application, and multi-agent coordination that achieve SOTA performance.

• Established history of open-source contributions pertaining to agentic systems or tool utilization (including code, datasets, or models) on platforms like GitHub or Hugging Face.

• Publications in prominent AI conferences (e.g., NeurIPS, ICML, ICLR, ACL, CVPR, ECCV).

🏝️ Benefits

• Flexible work arrangements

• Professional development opportunities