Technical Product Manager (Core Models)

Join Graphyte as a Technical Product Manager for Core Models, where you will own the product direction for the models that power our risk intelligence. You will work closely with Product, Data Science, and Engineering to prioritize model initiatives, define objectives, guide evaluation, and drive delivery. You will also serve as a customer-facing subject matter expert and coordinate with Model Risk Management processes. This role offers meaningful equity, performance reviews, collaboration across departments, and flexible work options.

• Own the Core Models product strategy and roadmap, prioritizing model build and refresh initiatives based on customer needs and business value.
• Translate use cases into clear model objectives, defining prediction targets and performance requirements that reflect how customers use the results.
• Drive model initiatives through delivery by defining scope, requirements, acceptance criteria, and release priorities; aligning Data Science and Engineering on execution.

• You are a highly analytical professional that operates at the interface of Product Management and Data Science
• You are rigorous and analytical, but you move with urgency and make practical decisions with imperfect information
• BS or MS in computer science, statistics, mathematics, engineering, or a related technical field, or bring equivalent practical experience
• You understand that a model’s value depends only on how well it solves a real customer problem
• 2+ years of product management, data science, and/or data analytics experience working with software powered by machine learning models or with significant model-driven capabilities
• You thrive by setting ambitious goals and tracking progress until they are met
• You are drawn to the intersection of product strategy and applied machine learning
• You understand agile software development and can translate priorities into clear requirements and an actionable backlog
• You communicate clearly with technical and nontechnical audiences and can explain model capabilities and limitations without overstating what the evidence supports
• You know how to turn customer needs into model requirements, evaluate whether an improvement matters in practice, and prioritize new capabilities alongside improvements to existing models
• You understand how label quality, class imbalance, and the representativeness of evaluation data affect conclusions about model performance
• You feel strongly about being mission-driven in your work, and above all else you are innately curious and motivated to learn new things
• You are collaborative by nature and have achieved success as a team player
• You are comfortable discussing precision, recall, and ROC analysis with data scientists, and explaining what those metrics mean to customers
• You can translate a use case into a clear prediction target, define what constitutes a true positive, and reason through the tradeoffs between missed risk and unnecessary alerts
• You have a strong working command of classifier performance metrics, including precision, recall, false-positive and false-negative rates, and ROC analysis and can interpret evaluation results, explain threshold tradeoffs, and connect those tradeoffs to customer outcomes
• You are comfortable partnering closely with Data Science and Engineering, distinguishing product decisions from modeling implementation choices while contributing meaningfully to both discussions
• You have an ownership mentality, thrive in ambiguity, and build alignment by clearly communicating the “why” and driving toward measurable outcomes
• Bonus: You bring familiarity with natural language processing, experience with financial crime or risk intelligence products, or exposure to model validation and Model Risk Management in regulated institutions
• You can turn an ambiguous customer problem into a well-defined model use case, prediction target, and measurable acceptance criteria