AI Workflow Engineer

Job Description

Requirements
- -----------

### Must have:

### - We bring at least 2 years of experience in backend engineering, AI automation, or integrating complex systems. - We have a proven record of building and supporting production-ready software with clear modular separation, such as distinct services for parsing, logic handling, and reporting. - We are highly proficient in at least two major programming languages, such as Python, JavaScript, Go, or Java, and we have experience working with SQL databases. - We have practical experience developing in live, non-simulated environments and managing multi-step system interactions. - We demonstrate exceptional attention to detail and can deliver concise, high-value technical feedback on complex system behavior. - We have experience designing multi-stage coordination workflows where data collection leads to reasoned outcomes. - We are comfortable connecting agents to real-world tools such as Supabase, Gmail, and other APIs to solve practical problems. - We understand how to implement persistent state and session tracking using MEMORY.md to monitor agent progress. - We can recognize subtle issues such as privacy exposure, authority escalation, and indirect prompt injection.

Responsibilities:
- ----------------

- We help train advanced generative systems by providing high-quality human feedback. - We contribute to the development of large language models that act as proactive, multi-step agents. - We design, coordinate, and refine complex architectural workflows for real-world use cases. - We evaluate system behavior in live environments and assess how agents perform across multiple turns. - We deliver precise technical observations that help improve orchestration, reliability, and overall model performance. - We work on projects that involve integrating data sources, reasoning over collected information, and producing structured outputs. - We identify vulnerabilities and failure modes, including privacy risks, improper permission handling, and prompt injection attempts.
- -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------

Company:
- -------

We are Outlier, and we help some of the worlds most innovative companies improve their AI agents through human feedback. Our work centers on training large language models to function as proactive, multi-step agents that can design, coordinate, and optimize complex real-world workflows, especially in architectural contexts. We offer the opportunity to contribute to cutting-edge generative AI projects alongside leading organizations and to help shape the future of autonomous agents.
- -----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------