Research Engineer, Codex
Palo Alto, hybrid
190,000–245,000USD / year
Role Summary
The Research Engineer builds and maintains Codex, Zettabyte's internal research and experimentation framework for GPU scheduling and optimisation. The role turns research ideas into reproducible experiments and carries the results into production systems.
Key Responsibilities
- Codex framework: build and maintain Codex, the internal research and experimentation framework.
- Abstractions: design abstractions for testing GPU scheduling, orchestration and optimisation ideas.
- Simulation and replay: implement simulation and replay environments for AI workloads.
- Rapid experiments: enable rapid experimentation across hardware, software and workload configurations.
- Research support: work with research scientists to operationalise experimental models.
- Reproducibility: make sure Codex outputs are reproducible, measurable and scalable.
- Production impact: integrate Codex insights into production systems and customer-facing platforms.
- Platform health: maintain the performance, reliability and extensibility of Codex.
Required Qualifications
- 5+ years of experience in systems engineering or platform development
- Strong software design skills and systems thinking
- Experience building internal tooling or experimentation frameworks
- Comfort owning complex, foundational infrastructure
- Strong collaboration across research and engineering teams
Bonus Qualifications
- Experience with simulation, emulation or workload replay systems
- Familiarity with GPU scheduling or orchestration
- Experience designing APIs for internal platforms
- Background in performance modelling or systems research
What Success Looks Like (First 90 to 180 Days)
- Codex supports a full experiment cycle from idea to measured result
- Simulation or replay environments in use by research scientists
- At least one Codex insight carried into a production system
- Codex is documented and reliable enough for others to extend
Why This Role Matters
- Ideas to evidence: Codex turns research ideas into measured results.
- Speed of learning: this role sets how fast Zettabyte can test and ship GPU optimisations.
Working Norms at Zettabyte
- Bias to action: deliver drafts quickly, then iterate on feedback
- Be precise: double-check numbers and facts; sloppy analysis creates downstream cost
- Make it usable: package outputs for decision-making, not data dumps
- Own the process: whoever drives a deliverable manages stakeholders and timelines
- Learn the ecosystem: understand GPUs, cluster operations and the buyer landscape