Member of Engineering (Reinforcement Learning (AI))

TL;DR

Member of Engineering (Reinforcement Learning (AI)): Improving reasoning and coding abilities of Large Language Models through reinforcement learning with an accent on designing and scaling RL environments and training algorithms. Focus on pushing the frontier of foundational models, implementing scalable RL pipelines, and diagnosing training instabilities.

Location: Remote (EMEA/East Coast)

Company

hirify.global is an AI research and engineering company building AGI to empower economically valuable work and scientific progress.

What you will do

• Research and experiment on improving reasoning and code generation for LLMs, owning the full experiment lifecycle.
• Translate state-of-the-art research in RL and LLMs into clean, reusable codebases.
• Design, analyze, and iterate on data generation and LLM training.
• Implement and scale RL training pipelines across multiple domains.
• Diagnose training instabilities and failures, proposing effective mitigation methods.
• Develop high-quality, reproducible, and maintainable code.

Requirements

• Deep understanding of Transformer architecture and scaling laws.
• Experience with mid-training and post-training techniques for reasoning or agentic models.
• Solid grasp of Reinforcement Learning concepts and experience with distributed, large-scale RL pipelines.
• Scientific publications in RL, LLMs, or reasoning models.
• Strong programming skills in Python and familiarity with PyTorch or JAX.
• Must be based in EMEA or US East Coast

Culture & Benefits

• Fully remote work with flexible hours.
• 37 days of vacation and holidays per year.
• Health insurance allowance for employees and dependents.
• Company-provided equipment and home office allowances.
• Wellbeing and learning budgets.
• Frequent team get-togethers and a people-first culture.

Hiring process

• Introductory call with a Founding Engineer.
• Technical interviews with Founding Engineers.
• Team fit discussion with the People team.
• Final interview with a Founding Engineer.