Software Engineer

About the role You'll sit at the intersection of Poolside's research, infrastructure, and product teams, acting as the connective tissue that brings agents into every corner of the Model Factory. The Agent Experience team owns the harness, skills, and tooling behind autonomous research and the daily experience of the researchers running it. In a given week, you might ship an agent that runs its own ablations end to end, build the distribution layer that lets a skill written by one researcher land in everyone's workflow the same day, or instrument trajectories so we can actually see where agents earn their keep. If you want to own the surface every researcher touches, work alongside the people building frontier models, and help shape what agentic research looks like next, this is the role for you. Your mission To establish, improve and accelerate self-improvement loops of the agentic connective tissue across all engineering and research teams at Poolside. Responsibilities - Partner closely with researchers (pre-, post-training), eval teams, infrastructure teams, and experiment platform engineers to translate research workflows into self-improving agentic flows. - Design and build robust backend services primarily using Python and Go to power our agents. - Drive adoption of agent-native research workflows. - Build out evals and datasets relevant for measure self-improvement capabilities of our own models and drive the inner flywheel of the model factory. - Integrate with existing research infrastructure and build dashboards to track adoption and usage metrics. - Work closely with our agent harness and runtime team to define the roadmap Skills & experience - Strong engineering fundamentals across backend architecture, API and database design, and operational excellence. - Demonstrated ability to lead ambiguous, cross-functional initiatives, aligning stakeholders around a shared technical vision. - Proven experience building agentic systems, skills, self-improvement loops or experience in non-trivial harness engineering. - Nice to have: Understanding of ML research workflows, including experiments, evals, training runs, evaluations, checkpoints, configurations, datasets, and research iteration cycles. - Nice to have: Experience in systems used in model training like W&B, Neptune, Dagster