Lead Machine Learning Engineer (Agentic Workflows)

BUILD SOMETHING BIGGER THAN YOURSELF 

 The Walt Disney Company, together with its subsidiaries and affiliates, is a leading diversified international family entertainment and media enterprise with the following business segments: Disney Entertainment, ESPN, Disney Parks, and Experiences and Products. From humble beginnings as a cartoon studio in the 1920s to its preeminent name in the entertainment industry today, Disney proudly continues its legacy of creating world-class stories and experiences for every member of the family. Disney’s stories, characters and experiences reach consumers and guests from every corner of the globe. With operations in more than 40 countries, our employees and cast members work together to create entertainment experiences that are both universally and locally cherished. 

How you will contribute to our magical experience?  

Disney Financial Insights (D-Fi) is a new platform reimagining how Finance teams and budget owners across Disney interact with financial data — replacing manual pulls, static reports, and email-driven approvals with one AI-powered, orchestrated experience built on top of Disney financial systems. 

We're hiring the founding engineering team for this AI framework. One theme unifies every engineering role on D-Fi: autonomy and trust. You build the agents that act autonomously — designing how they execute finance processes end-to-end and knowing exactly when to hand off to a human instead. You'll own the agentic workflow layer: the core of D-Fi's autonomous finance capability. This means designing, building, and maintaining the multi-agent orchestration framework that enables touchless execution of real finance business processes. 

The Lead Machine Learning Engineer (Agentic Workflows) will be responsible for:  

• Own the agentic workflow layer. Design, build, and maintain the multi-agent orchestration framework that enables touchless execution of finance business processes — close cycle automation, variance analysis, forecast change routing, report generation, and anomaly response. 
• Own agent architecture decisions. Determine how agents are structured, how they hand off to each other, how they handle failure modes and exceptions, and how human-in-the-loop escalation is triggered when autonomous action isn't appropriate. 
• Consume and enforce delegated identity in agent execution. Ensure every agent action executes under the identity and entitlements of the user (or process) it's acting on behalf of — propagating that identity context correctly through every agent-to-agent handoff, and never allowing an agent to execute an action the underlying human wouldn't be authorized to take directly. 
• Set the prompt and context engineering standard for the team. Lead the discipline of how LLMs are instructed, constrained, and evaluated across every agent on the platform, not just your own. 
• Partner directly with finance domain experts. Map real finance processes into agentic workflows that can run with minimal human intervention, translating process knowledge into system behavior. 
• Design for graceful failure, not just success. Build exception handling and escalation paths so that when autonomous action isn't safe or appropriate, the system routes to a human cleanly and traceably. 

What we need from you to be part of this story: 

• 7+ years of software or ML engineering experience, with demonstrated ownership of complex systems end-to-end. 
• Advanced level of English (must). 
• Bachelor's degree in Computer Science, Engineering, or related field; advanced degree a plus but not required. 
• Hands-on experience designing and shipping multi-agent or agentic LLM systems in production — not prototypes or demos. 
• Strong judgment for failure-mode design: knowing where autonomy should stop and human review should start. 
• Comfort setting technical standards other engineers follow, even without formal management authority. 
• Genuine curiosity about finance processes — you don't need a finance background, but you'll need to learn how forecasting, variance analysis, and approval workflows actually work. 

You'll be architecting, designing and writing production code. 

AI/LLM: 

• Production experience building multi-agent systems — agent-to-agent handoff, task decomposition, and orchestration patterns. 
• Strong prompt and context engineering skills, including designing constraints and guardrails for autonomous LLM-driven actions. 
• Experience with agent orchestration frameworks (e.g., LangGraph, AutoGen, CrewAI, or equivalent) and major model providers/APIs (Anthropic Claude, OpenAI, or equivalent). 
• Understanding of evaluation methodology for agentic systems — task success rate, failure mode analysis, and regression testing as prompts and workflows evolve. 
• Experience designing human-in-the-loop escalation patterns for AI systems operating with partial autonomy. 

Data & Databases: 

• Working SQL fluency and comfort querying a modern cloud data warehouse (Snowflake preferred) to ground agent decisions in real data. 
• Understanding of how agents should consume data from a governed semantic layer rather than querying raw sources directly. 

Programming Languages & Engineering: 

• Python fluency required — this is the primary language for agent and orchestration logic. 
• Experience with event-driven or workflow orchestration systems (e.g., Temporal, Airflow, or equivalent) is a strong plus. 
• Familiarity with testing strategies for non-deterministic systems (LLM outputs, agent decisions). 
• Experience with cloud infrastructure (AWS, Azure, or GCP) sufficient to deploy and scale agentic services. 

Identity & Access Management (IAM): 

• Understanding of delegated authorization patterns (token exchange, on-behalf-of/OBO flows) sufficient to correctly consume and propagate a user's identity context through multi-agent workflows. 
• Experience designing systems that enforce least-privilege execution — ensuring an agent's actions never exceed the entitlements of the human or process it's acting for. 
• Familiarity with JWT-based identity tokens and claims, and how to thread identity context through asynchronous, multi-step agent execution without losing or mutating it. 
• Comfort building audit trails that tie every autonomous action back to the originating user identity, for traceability and compliance. 

Enterprise Systems (plus, not required): 

• Exposure to ERP/EPM platforms such as SAP or Oracle EPM/Cognos, or experience integrating AI systems against enterprise business processes. 

Additional Information:  

At Disney, we believe it takes great people to create the memorable experiences, products and services our consumers know and love. We are a team of storytellers and hire under equal opportunity regardless of race, color, ancestry, religion, sex, national origin, sexual orientation, age, disability and gender identity. We foster a corporate culture of belonging and inclusiveness, which empowers our staff and company to thrive. At Disney, everyone has the opportunity to speak up and participate, grow and share ideas. 

“You can design and create, and build the most wonderful place in the world. But it takes people to make the dream a reality.” Walt Disney 

This position is with TWDC Argentina S.A., which is part of a business we call The Walt Disney Company (Corporate).

To learn more about Disney's guidelines on the use of AI for candidates, click here.

Having technical issues? View the FAQs for help.