Founding Machine Learning Engineer

Engineering/San Jose, CA/Full-time Founding Machine Learning Engineer You'll design and deploy agentic AI systems that reason about patentability, prior art, and claim construction - problems that have stumped pure foundation models. This means training proprietary models from scratch, building multi-agent workflows that coordinate specialized classifiers and reasoning systems, and creating evaluation frameworks grounded in real patent prosecution standards. You'll go deep into both the ML engineering (distillation, fine-tuning, distributed training) and the patent domain itself, learning to recognize what makes an invention truly novel. What You'll Do —Design and build agentic systems that coordinate multiple models to accomplish complex tasks—Work with professionals to create evaluation datasets and judgement frameworks for agent performance—Orchestrate agents that leverage specialized models for classification, prediction, and reasoning tasks—Post-train and fine-tune open source models to work effectively within agent architectures—Figure out scalable deployments for agent systems and their underlying models—Build continuous pipelines to improve agents and retrain models as the landscape changes What We're Looking For —Have at least 3+ years of ML research/applied ML experience in one of: CV, RL, Graph ML, NLP (ML based approaches not just statistical)—Have experience building agentic systems, multi-agent workflows, or LLM orchestration frameworks—Have experience deploying ML models on distributed compute/cloud compute—Have experience with model distillation/getting maximum performance from minimal compute (we will have the budget for training and hosting bigger models but preferably due to the ability to break down tasks into smaller and smaller subtasks, we would prefer deployments of smaller models) Nice to Have —Experience with agent orchestration and evaluation frameworks—Distillation experience—Understanding of implementing LORA + QLORA systems Apply for this role Form not loading? Open it in a new tab