Senior Staff Privacy Engineer - Moveworks

Our AI products rely on sensitive data to enable customers to get work done. With this in mind, privacy is a crucial part of ensuring that data is handled properly throughout the workflow, from what we collect and how we use it to the controls we apply across models, tools, logs and data pipelines. In this role, you will be a key privacy engineer, writing code and shaping the architecture and protections across our products. You will improve sensitive data detection and masking, assess the privacy risks of ML models, and own the implementation of customer data-use and retention controls. You will work with our Privacy, ML, Security, Product and Infrastructure teams from design through production. This is a Senior Staff individual contributor role. You will be hands on, setting technical direction across teams and writing production code. What you get to do in this role: • Assess, benchmark and select ML models and techniques that meet our privacy requirements. Build representative evaluation datasets and compare detection coverage, false positives, latency, cost and the impact on product quality. • Own the implementation of those models in our products, from prototype through production. Train or fine-tune where needed, build the surrounding services and APIs, and optimize inference performance and reliability. • Build and deploy sensitive data detection, masking, redaction and tokenization for structured and unstructured data across logs, traces, APIs, analytics and annotation tools. Implement consistent controls for when data is masked and who can access the unmasked content. • Build privacy evaluations for ML models and the systems around them, including memorization, training-data extraction, membership inference and sensitive data disclosure through retrieval or tool use. Implement mitigations and add regression tests to model and product release workflows. • Write and maintain code that enforces customer requirements for data use, access, retention and deletion across services and data stores, including the distinction between permitted inference, evaluation, annotation and training. • Prototype and implement privacy-preserving approaches to evaluation and synthetic data. Measure privacy risk and utility, validate the assumptions behind the approach, and integrate suitable solutions into our ML pipelines. • Conduct privacy design reviews with engineering, identify gaps in proposed data flows and architectures, and own or partner on the code changes needed to address them. Build reusable components and engineering standards that help teams apply these protections consistently. • Monitor privacy controls and models in production, investigate failures, and ship improvements. Use error analysis and production feedback to improve detection quality, address regressions and close customer-reported privacy issues. What success looks like in your first year: • ML models selected and deployed for priority privacy use cases, with measured improvements in detection quality and acceptable production latency and cost. • Shared masking and data protection components integrated into the highest-priority product surfaces, with automated tests for coverage and access behavior. • Repeatable privacy evaluations running as part of model and product releases, with findings translated into shipped fixes. • Priority data-use, retention and deletion requirements implemented and verified in production, with monitoring that identifies failures.   To be successful in this role you have: • Typically 7+ or more years building production software, including substantial experience in privacy engineering and applied ML. • Strong Python skills and experience with Go, Java, C++ or another production systems language. You have shipped, tested