*software engineer, infrastructure & devops*
*train large-language models (llms) to write production-grade infrastructure and devops code*
help teach ai how to write, debug, and optimize infrastructure code like a top-tier devops engineer.
- compare and rank terraform or iac code snippets, explaining which is more reliable, efficient, or scalable.
- refactor or repair ai-generated infrastructure setups for correctness, security, and clarity.
- provide structured feedback (edits, test outcomes, architectural notes) that feeds into the rlhf pipeline.
- * end result*: the model learns to reason about devops the way _you_ do—smart, scalable, and safe.
*rlhf in one line*
generate iac or backend code ? Expert engineers rank, edit, and explain ? Convert feedback into reward signals ? Reinforcement learning tunes the model to think like a real infra engineer.
*what is needed*
- * 4+ years of professional software-engineering experience*, ideally in backend, infrastructure, or devops.
- * extreme attention to detail and excellent written communication*—you'll be writing a lot of thoughtful code reviews and architectural justifications.
- * fluency with cloud infrastructure*, especially *aws*.
- * hands-on experience with terraform* and infrastructure as code practices.
- strong instincts for *performance, cost optimization, and security*.
- comfortable in a *low-oversight, async-first remote environment*.
- you enjoy reading documentation and specs—seriously.
*what is not needed*
- no prior rlhf or ai-training experience required.
- no machine learning expertise needed—just deep infra/devops experience and clear thinking.
*tech stack*
we are especially looking for backend engineers working in infra or terraform-heavy environments.
experience with any of the following is highly valued:
- * terraform, aws, gcp, or other cloud providers*
- * node.js or backend scripting*
- * ci/cd pipelines*
- * infrastructure as code (iac) best practices*
*logistics*
- * location*: fully remote (work from anywhere)
- * hours*: minimum 15 hrs/week, can go up to 40 hrs/week
- * engagement*: * contract