Developer, Agent Simulation/Evaluation Pipelines (Personal/Avrello Tech)
Developed and maintained data simulation pipelines for agent training and evaluation workflows. Built a Python-based web application simulation pipeline focused on benchmarking AI agent workflows, including state capture and action sequencing on real software environments. Enabled reproducible evaluation and training of computer-use agents for platform features against real or simulated data patterns. • Evaluated agent interactions and outcomes using systematic benchmarking. • Labeled and assessed agent actions based on code output and environment state changes. • Built robust tools to automate simulation data capture and task completion verification. • Focused on improving training data quality for AI agent pipelines.
2024 - 2025