Member of Technical Staff, Alignment

Sourced from feeny.ai

Abundant is an applied research lab scaling reinforcement learning for safe and reliable agentic capabilities. As a Member of Technical Staff on Alignment, you will ensure alignment is embedded in deployed code, runbooks, and published research. The role combines hands-on engineering with safety research on reward hacking, monitoring, red teaming, and agent containment.

• Run experiments on how environments shape agent behavior, including reward hacking, task cheating, unsafe tool use, and deception over long runs
• Break and harden graders, turning discovered attacks into shared regression tests
• Build monitoring tools, red team harnesses, and safety benchmarks for real agent runs
• Secure environments through network isolation, escape prevention, and credential controls
• Pursue and publish research on oversight, control, or long-horizon autonomy
• Define safety boundaries and communicate clearly what partners can and cannot safely use

• Strong engineering background with ability to build test pipelines and navigate unfamiliar codebases
• In-depth knowledge of LLM safety: attacks, red teaming, monitoring and control evaluations, reward hacking
• Ability to turn vague model concerns into rigorous experiments and interpret results
• Close reading of transcripts to catch subtle errors
• Experience putting research into production systems, especially post-training, distillation, or evaluation
• Clear written communication that influences decisions
• Comfort making fast decisions with incomplete evidence

• Published work on control, dangerous capability evaluations, oversight, or interpretability
• RLHF or RLAIF experience
• Built or ran a public benchmark or agent task suite
• Experience running many agents at once with sandboxes, containers, and logging
• Advised on AI safety or governance policy