6 hours ago
Remote, United StatesSenior
Base Salary
$176k - $243k/yr
Responsibilities
- Design, build, and deploy high-throughput Reinforcement Learning from Verifiable Rewards environments for evaluating LLM action sequences.
- Develop automated test harnesses, sandboxes, verification engines, deterministic reward functions, telemetry, and trajectory logging for cybersecurity tasks.
- Integrate Bugcrowd’s Mayhem platform and vulnerability feeds into scalable RL environment-generation pipelines.
- Build safe, isolated, reproducible execution environments capable of supporting thousands of simultaneous exploitation and patching trajectories.
- Create low-level instrumentation and debugging tools for monitoring memory states, processes, and network behavior.
- Optimize environment reset latency and infrastructure performance for massive-scale parallel sampling and distributed RL training.
- Benchmark frontier AI models on fuzzing, exploit generation, patch validation, and other offensive and defensive security challenges.
- Collaborate with frontier AI researchers to define benchmark formats, observation spaces, and verifiable evaluation metrics.
Requirements
- Understand RL training workflows for modern LLM systems, including execution-based feedback and RLVR.
- Develop applications in Python and low-level systems software in C; Rust experience is a strong plus.
- Understand software vulnerabilities, binary exploitation, fuzzing methodologies, or program analysis.
- Have experience with DevOps pipelines, reproducible builds, Linux systems, and low-level debugging.
- Have experience working with or building benchmark environments such as CTFs, SWE-bench, security challenges, or execution sandboxes.
- Preferred qualifications include experience with custom reward functions, ground-truth verifiers, automated grading engines, AI safety or reasoning-model evaluation, sanitizers such as ASan/MSan, compiler instrumentation, automated exploit generation, cybersecurity benchmarks, CTFs, or open-source AI evaluation frameworks.
Benefits
- Fully remote, work-from-home 100% of the time.
- Eligible positions may participate in a discretionary bonus program or commission plan.
- Reasonable accommodations are available for qualified individuals with disabilities.
- Employment is subject to applicable background-check procedures.
Categories
About Bugcrowd
Bugcrowd builds a crowdsourced security platform used by security teams to run bug bounty, vulnerability disclosure, and penetration testing programs. The service combines SaaS workflow and triage with access to a vetted hacker community, paying researchers per validated finding while customers subscribe to managed programs. Founded in 2012 and headquartered in San Francisco, the privately held company emphasizes AI-driven matching and signal processing to prioritize actionable vulnerabilities.