We're looking for AI-native, builder-type engineers who thrive in ambiguity.
l Experience architecting long-context, multi-step, or persistent agents
l Knowledge of agent planning architectures (ReAct, Reflexion, Tree-of-Thought,
Graph-of-Thought, RA-As-A-Graph, AutoGen, CrewAI, Swarm)
l Ability to design robust state machines for agent behavior
l Experience with long-term memory, episodic memory, vector databases
l Memory compression / summarization strategies
l Schema design for agent state
l Experience creating eval suites (MMLU-style, behavioral tests, golden datasets)
l Familiarity with output scoring + model routing
l Human-in-the-loop feedback pipelines
Familiarity with:
l Strong backend engineering (Python, FastAPI, Node, or Go)
l Kafka / event queues / async workers
l Docker, Kubernetes, CI/CD
Nice to have:
l Familiar with AI hardware architecture design, cloud architecture, AWS certificate is a strong
plus.
l Research ability to implement new agent architectures from papers.
Critical differentiator (please highlight this)
You have built AI agents yourself—either: as a successful side project / indie project, OR as a
startup founder / technical owner who designed and implemented agent systems end-to-end
Nice to Have: Hands-on model tuning, AI system building, open-source work, startup experience,
curiosity for recruiting / org design / people systems—or willingness to rapidly learn