Staff+ Research Engineer – Recursive Self-Improvement Lead
Sequen AI · Remote · United States ·
COMPENSATION
$350k – $400k
Listed on the employer's job board (Ashby)
- Staff+ level role
- Fully remote or hybrid
- Python + LLM agents
- 7+ years experience
- Applied research focus
You will lead the Recursive Self-Improvement team at Sequen, focusing on making their AI agents better at conducting ML research. This applied research position requires writing production-quality code and measuring success by the tangible improvements delivered to client models in real-world environments.
What you'll do
- Define the research roadmap for how the system improves its own research process
- Design exploration policies and compute allocation strategies for parallel agents
- Convert recorded research sessions into training and evaluation signals
- Build evaluation metrics to define and measure research quality
- Ensure knowledge gained in one session improves performance in subsequent tasks
- Translate promising academic ideas into measured experiments within weeks
- Collaborate with applied scientists to identify agent shortcomings in client work
What you'll need
- Seven or more years of experience in applied ML research or research engineering
- Hands-on work with LLM agents, AutoML, meta-learning, or recursive self-improvement
- Proven ability to take research prototypes to measured production improvements
- Strong experimental rigor including baselines, ablations, and reproducibility checks
- Proficiency in writing production-quality Python and building end-to-end experiments
- Deep understanding of frontier model behaviors in long agentic loops
- Experience leading or mentoring small research teams
Nice to have
- Peer-reviewed publications in top conferences like NeurIPS, ICML, or ICLR
- Experience with reinforcement learning, RLHF, or reward modelling
- Background in building evaluations or benchmarks for LLMs and agents
- Exposure to search, recommendation, or learning-to-rank domains
- Experience with multi-agent systems and agent swarms
Stack
- Python
- LLM agents
- AutoML
- Reinforcement Learning
How you'll work
- Flexible hybrid or remote configuration available
- Unlimited paid time off
- Early-stage team environment with high autonomy