New opportunity

Member of Technical Staff (Evals & Post-Training)

Ambral · New York; San Francisco, United States

About this role

Build environments, evals, and training systems that convert recorded enterprise data into replayable environments and eval sets; design graders, mine tasks and trajectories, and train and evaluate agents over long horizons while scaling evaluation and training runs for enterprise deployment.

Skills for this role

Reinforcement LearningLLM post-trainingEvaluation InfrastructureAgent EngineeringEnvironment DesignReward DesignContext EngineeringGrader DesignMining Tasks from Historical WorkflowsCreating Eval SetsModel EvaluationInference Cost OptimizationLong-horizon Agent TrainingObservabilityReproducibilityScalable Training SystemsProduction Software EngineeringProcessing LargeMessy DatasetsResearch-to-Production Implementation