Data Scientist

Overview In this role you will build and deploy production models that predict cycle time, tool life, quality, and demand for high-mix, low-volume aerospace parts. You’ll create geometry-informed features and embeddings to transfer behavior to new parts, validate models with time-aware backtests, and own end-to-end model pipelines on Hadrian’s platform. You’ll collaborate with ML Platform and Data Engineering to deploy, monitor, and retrain models, and translate predictions into actionable manufacturing decisions. This role sits at the intersection of forecasting, representation learning, and geometric modeling to accelerate factory throughput and reduce costs. Compensation / BenefitsMedical, dental, vision, and life insurance401kRelocation support may be providedFlexible vacation policyEquityOn-site location (Los Angeles, CA) ResponsibilitiesBuild and ship production models for cycle time, tool life, quality, and demand with calibrated uncertaintyEngine­er geometry-based features and geometric/graph models to predict cycle time, cost, DFM, tolerance risk, and triage probabilityDevelop part and operation embeddings representing geometry, material, tolerances, and route to enable cold-start transferValidate models with time-ordered backtesting and assess deep vs classical methodsOwn end-to-end model pipelines: training, serving, monitoring, retraining with ML Platform and Data EngineeringDetect drift and quality anomalies in production to continuously improve predictionsTurn predictions into productionDecisions for quoting, scheduling, capacity, and DFM; design experiments and A/B tests and hand off results to operations Key requirementsForecasting and prediction on real, messy manufacturing data with honest uncertaintyRepresentation learning and embeddings; similarity and retrieval; transfer/few-shot for sparse dataDeep learning with PyTorch and judgment to know when to use itStrong classical ML and statistics (GBMs, Bayesian/hierarchical, survival, causal)Validation with backtesting, leakage control, calibrationPython; feature engineering from messy processes to usable model inputsExperience deploying and monitoring models with attention to driftAbility to work with limited, high-value data and borrow strengthproblem-solving and curious mindsetability to communicate model insights to non-technical stakeholderscollaboration with cross-functional teamsPyTorchGeometric deep learning (mesh/point-cloud, GNNs)CAD/B-rep feature engineering