Research Engineer - Post-Training for Agentic Coding (AI)

TL;DR

Research Engineer - Post-Training for Agentic Coding (AI): Developing and implementing post-training solutions for enterprise-grade coding agents using static analysis and LLM techniques with an accent on reinforcement learning from verifiable rewards, fine-tuning, and safety alignment. Focus on designing experiments, iterating prototypes, and converting research into high-impact products for agentic software development.

On-site in London

Company

Leader in AI code review and verification, enabling reliable AI-generated code for Fortune 100 companies via hirify.globalQube and agentic tools.

What you will do

• Develop advanced post-training products for coding agents that generate enterprise-standard code.
• Collaborate with researchers and engineers to design experiments, build prototypes, and productionize successful ones.
• Contribute ideas in cross-disciplinary team to advance coding model post-training.
• Stay current with LLM and agentic developments, explain complex concepts to diverse audiences.

Requirements

• Master’s or PhD in Computer Science, Machine Learning, or related field
• Strong industry ML experience with modern software engineering practices
• Fluency in Python and core ML frameworks; Rust or C#/C++/JS/TS/Java a plus
• Expertise in post-training techniques: RL from verifiable rewards, GRPO, offline RL, PEFT, SFT, safety alignment
• Experience with large-scale data processing and cloud infrastructure (AWS, Databricks)
• Track record driving research to prototypes and products; excellent English communication

Culture & Benefits

• Global team across hubs in Austin, Bochum, Dubai, Geneva, London, Singapore, Tokyo, Washington D.C.
• CODE mindset: committed, quality-obsessed, deliberate, effective teamwork
• Fast-paced growth in profitable company building AI software revolution
• Diversity, equity, inclusion focus; equal opportunity employer

Hiring process

• Comprehensive background check and reference verification
• AI tools may assist in application review; final decisions by humans
• No agency submissions; contact for accommodations