About the job
Our team is building the next generation of AI-native risk intelligence systems to address emerging challenges driven by large-scale AIGC content production. As content creation becomes automated and adversaries adopt systematic experimentation (e.g., large-scale template variation and rapid iteration), traditional rule-based and single-model approaches are no longer sufficient. We are transitioning from monolithic LLM applications to a structured multi-agent architecture that emphasizes: Tool-augmented reasoning (ReAct-style systems) Modular skill composition Execution traceability and observability Feedback-driven system evolution Cross-domain risk reasoning We are seeking an experienced technical leader to define and implement this architecture.
Responsibilities
- Architect and Implement Agent Systems
Design structured agent workflows (e.g., Evaluate → Validate → Reflect → Summarize)
Implement ReAct-style tool allocation and reasoning frameworks
Develop short-term and long-term memory architectures
Ensure robustness under adversarial and evolving conditions
- Lead Multi-Agent Architecture Development
Design orchestration layers for coordinating vertical domain agents
Build modular Skill systems for extensibility and reuse
Define execution graph standards and planning abstractions
Establish traceability mechanisms for debugging and auditability
- Develop Open Risk Detection Capabilities
Architect systems capable of identifying previously unseen risk patterns
Implement execution trace–driven optimization loops
Translate feedback signals (FP/FN, reviewer overrides, drift signals) into system improvements
Enable proactive rather than purely reactive detection systems
- Establish Engineering Standards for Agent Systems
Define traceability, observability, and guardrail requirements
Evaluate and integrate multi-agent frameworks where appropriate
Ensure production-readiness, scalability, and reliability
- Provide Technical Leadership
Own the technical roadmap for Agent and multi-agent systems
Partner cross-functionally with Risk, Safety, Infra, and ML teams
Mentor engineers and drive architectural rigor
Qualifications
Minimum
- 5+ years of experience in software engineering or applied AI systems
- Deep understanding of LLM-based agent architectures (ReAct-style reasoning/Tool calling systems/Workflow orchestration/Memory design patterns)
- Experience designing distributed or modular AI systems
- Strong backend engineering skills (Python or equivalent)
- Experience operating systems in adversarial or high-stakes environments
Preferred
- Experience in trust & safety, risk detection, or adversarial ML
- Familiarity with multi-agent orchestration frameworks
- Experience building systems with execution trace logging and observability
- Background in RL-style policy optimization or iterative system refinement
- Demonstrated experience leading high-impact technical initiatives