Principal Machine Learning Engineer

Cushman & Wakefield is seeking a Principal Machine Learning Engineer to build and maintain AI and machine learning systems for its business. The role leads complex ML engineering, MLOps/LLMOps, AI governance, and architectural standards while embedding across teams and mentoring engineers.

Responsibilities:

• Design and develop agentic AI systems — including multi-agent orchestration, tool-use, evaluation, and autonomous task-execution pipelines — to automate complex enterprise workflows
• Enforce model development standards encompassing data preprocessing, feature engineering, model selection, hyperparameter tuning, evaluation, and documentation
• Partner with data engineering teams to ensure robust, scalable, and high-quality data pipelines that support model training and inference
• Implement the organization's MLOps and LLMOps capabilities, including CI/CD pipelines for data wrangling, feature engineering, model training, evaluation, deployment, and monitoring
• Help define and enforce standards for model versioning, experiment tracking, reproducibility, and model registry management
• Implement robust model monitoring frameworks to detect performance degradation, data drift, concept drift, and bias in production systems, with automated alerting and retraining triggers
• Implement AI governance frameworks, including policies for model risk management, fairness, explainability, privacy, and security
• Champion sound AI principles across the global business, ensuring that human oversight and accountability are embedded in every stage of the AI development lifecycle

Requirements:

• Demonstrable experience shaping the end-to-end architecture, development, and deployment of AI, including machine learning, GenAI, and Agentic models that are tailored to business use cases
• 10+ years building software or machine learning systems, including 6+ years shipping and operating machine learning in production — served, monitored, and maintained after launch
• Prior experience as the senior-most or near-senior-most engineer on a production ML or data platform, with responsibility for architectural decisions others had to live with
• A track record of setting technical standards that outlasted your own involvement, and of leading through influence without formal authority
• Strong hands-on Python and SQL, and the software engineering fundamentals that keep ML systems maintainable at scale — version control, automated testing, code review, modular design — plus the discipline to enforce them across teams
• Expert hands-on proficiency with Databricks (Spark, Lakeflow Declarative Pipelines, Delta Lake, Lakebase/Postgres, Unity Catalog, Model Serving) and the Azure data ecosystem
• Deep experience on a comparable lakehouse platform will be considered, with the expectation of becoming effective on Databricks quickly
• Experience with MLflow or an equivalent for experiment tracking, model registry, tracing, and reproducibility
• At least one year building agentic systems that reached production use, or equivalent demonstrated work
• Proven ability to mentor engineers at every level and materially raise the technical standard of the teams you work with
• Familiarity with modern machine learning frameworks and a strong instinct for reusable, scalable engineering patterns
• Experience establishing an ML or AI platform capability in an organisation that did not previously have one
• Experience embedding across multiple teams, geographies and time zones as a principal-level technical contributor
• Familiarity with CI/CD and infrastructure-as-code tooling for data pipelines using Azure DevOps, Databricks Declarative Automation Bundles (DABs), GitHub Actions, Terraform, or equivalent
• Good understanding of data governance, metadata management, and cataloguing in enterprise environments
• Exposure to model serving and inference at scale — containers, REST endpoints, latency and throughput targets — or to feature stores, retrieval pipelines, or LLM cost-attribution tooling
• Model Context Protocol (MCP) or comparable tool-integration standards
• Commercial real estate, professional services, or another domain where the data is messy, entities are inconsistently defined, and subject matter experts are the source of truth