Senior Manager, Application & Machine Learning Engineer

Senior Manager, Application & Machine Learning Engineer
Role Overview
You will build the models, optimizers and agents behind pricing, pack and promotion decisions across Coca-Cola's markets - from XGBoost and Gurobi to LangGraph - on Databricks and Azure, working with some of the largest commercial datasets in consumer goods.
The role sits in the Product Engineering team of Coca-Cola's Global Digital Network, in a small, senior team that turns business questions from operating units and bottlers into secure, data-driven agentic solutions with executive-ready insight, increasingly delivered inside Microsoft Teams.
This is a hands-on senior engineering role, not a people-management one. Most of your week is spent building and shipping - writing Python, training and scoring models, engineering prompts, building agents - and what you ship is used by the people who run the business.
What You'll Do

• Build and ship production machine learning, optimization and forecasting models on Databricks and Azure - on very large structured and unstructured datasets - that feed commercial decisions on pricing, packs and promotions.
• Design and deliver generative AI and agentic solutions - RAG, tool-calling agents and multi-agent workflows on frameworks such as LangGraph, Semantic Kernel or Azure AI Agent Service - that give a category or brand manager a grounded, auditable answer in Teams.
• Work directly with product managers, data engineers, architects and business leaders in operating units and bottlers to turn an ambiguous question into a model, an optimization and an agent that explains its answer.
• Own your solutions in production: evaluation of model and LLM output, monitoring, MLOps and lifecycle management, and compliance with enterprise security, privacy and responsible AI standards.
• Set the engineering patterns the team reuses - code quality, testing, CI/CD, observability - so that delivery gets faster and safer with each release.
• Help shape a team that is being built now: your choices of algorithms, frameworks and practices become the standard.

What You'll Bring
Required Skills:

• 7+ years in software, application or machine learning engineering or data science, including 3+ years delivering machine learning or AI solutions in production.
• Python at production quality: you write code that others ship and maintain.
• Applied machine learning and data science on very large datasets: model development, deployment and MLOps.
• One to two years of demonstrable prompt engineering, and at least one generative AI application delivered to production with real users.
• At least one agent built and shipped on a recognized framework (LangGraph, Semantic Kernel, AutoGen, OpenAI Agents SDK or Azure AI Agent Service), including retrieval-augmented generation.
• Production software practices: APIs and services, CI/CD, automated testing and observability.

Preferred Skills:

• Hands-on Microsoft Azure/AWS Databricks, including Spark / PySpark and Azure data services.

• XGBoost, forecasting and simulation models.
• Optimization modeling with Gurobi or open-source equivalents, and operations research techniques.
• Multi-agent architecture and LLM evaluation frameworks.
• Pricing, promotion, profitability, forecasting or supply chain use cases, ideally in consumer goods or for a global organization.
• R, in addition to Python.

Why Join

• Scale that is hard to find: terabyte-scale commercial data from markets on four continents, and decisions that move volume and profit in the field.
• A modern stack and the mandate to use it
• A small senior team being built now, with three roles opening together: you set the standards rather than inherit them.
• Direct access to the business leaders whose decisions your work informs, and the backing to say what is and is not technically feasible.

What Success Looks Like
In your first six months, a model or agent you built is used by an operating unit or bottler in a live decision cycle, with its output evaluated and monitored in production.
By twelve months, you own the optimization layer behind one of our pricing or promotion products and the agent that explains its answers to the people acting on them - and the patterns you set are reused across the team.