At Schnell AI, I experimented with cutting-edge machine learning models and agent frameworks to evaluate performance across candidate architectures and hyperparameter settings. I designed, built, and iterated on on-premise AI pipelines, ensuring reliable data ingestion, model training, and inference workflows. I also prototyped agentic workflows for regulatory tasks, accelerating policy-compliance checks and streamlining manual review processes. Collaborating closely with the founding team, I helped define requirements and develop internal tools and product features that support core business needs, while documenting technical processes and delivering presentations to guide stakeholders and inform decision-making.