Sr. Manager Data Science

Here at Scotts Miracle-Gro there is no such thing as a typical day. Our culture is constantly energized by new and exciting growth opportunities and at a rapid pace.

Below are details on an open job. The role in one line:
Lead a data science team that turns Scotts' commercial questions into production ML and analytics, pricing and elasticities, demand and POS forecasting, category and audience insight, and the models that power our AI agents, using agentic practices to move faster while holding a high bar for ML rigor and engineering discipline.

Your models ship, they don't die in notebooks. Data Science sits inside the same org as the build and agent engine, so the work goes into production agents and applications, not slide decks. Agentic-first, with judgment where it counts. The team uses AI agents and coding assistants to absorb the formulaic ~45% of data science work (profiling, EDA, feature scaffolding, hyperparameter search, monitoring checks) so people spend their time on method, interpretation, and business impact.

You focus on modeling and impact, not plumbing. A dedicated Data Engineering function owns the data foundation (pipelines, ingestion, the lakehouse), so your team builds on solid ground.

• price and promotion elasticities
• POS and demand forecasting
• category and market analysis
• audience and activation analytics
• the models that feed our AI agents

• lead, coach, and grow a group of data scientists and senior analysts
• set technical standards
• hire for the net-new skills as the function scales

• a mature, reproducible ML lifecycle across the team, from experiment to production to monitoring to retirement

• a clear, measurable connection between the team's models and outcomes (forecast accuracy, margin, conversion, revenue), and the ability to tell that story to non-technical partners

• Frame ambiguous business questions as tractable modeling problems; choose the right method and know its limits
• Deliver models across the relevant families: time-series forecasting, causal and elasticity modeling, regression and classification, gradient-boosted trees, and modern ML as appropriate
• Set the standard for evaluation: define success metrics and golden datasets up front, and hold models (and agent-assisted analysis) to them. Eval-driven development is the default

• Put AI coding and analysis agents (for example Cursor, Claude Code, and notebook or pipeline agents) into the team's daily workflow to automate repetitive work and compress cycle time
• Build agent-assisted workflows for EDA, data profiling, feature engineering, hyperparameter search, and monitoring, with human review at the decision points
• Apply sound judgment on where to trust a model and where to ground or verify it; teach the team to do the same

• Treat models as production software: reproducible pipelines, version control, testing, and clean, reviewable code
• Own CI/CD for ML, model and data versioning, lineage tracking, staged rollouts, drift and performance monitoring, retraining triggers, rollback, and model governance
• Package models as services and APIs so they integrate cleanly into agents and applications

• Coach and develop the team; recruit and level talent; set a culture of rigor, speed, and continuous learning
• Sequence work with business Product Owners; manage dependencies with Data Engineering and the build teams
• Communicate impact and tradeoffs clearly to technical and business audiences

• Strong ML foundations. Solid grounding in ML algorithms and statistics, able to select, tune, and critique methods (forecasting, causal/elasticity modeling, boosting, classical ML), not just call libraries
• ML engineering. Production-quality Python; reproducible pipelines; fluent with Git, testing, containers, and APIs. MLOps, CI/CD, and versioning. Hands-on experience operating a mature ML lifecycle: CI/CD for ML, model and data versioning and lineage, monitoring, retraining, rollback, and governance at scale
• Agentic fluency. Confident daily use of AI coding and analysis tools; working understanding of LLM evaluation, RAG, embeddings, vector search, and agent workflows, including their failure modes
• People leadership. Track record leading and growing a data science team, coaching individuals, and prioritizing across competing stakeholders
• Business acumen. Demonstrated…