• Own the full machine learning lifecycle from research to production • Leverage Computer Vision and NLP techniques (e.g. BERT, ViT) to extract insights from unstructured data • Advocate for sustainable tradeoffs, architecting solutions that balance product needs and long term goals • Design experiments to quantify the business impact of machine learning models • Collaborate with other MLEs on tooling improvements for model training and deployment • Write robust, production-grade Python code to productionalize feature engineering, model training, and deployment • Participate in the on-call rotation to assist engineers in troubleshooting

• Opportunity discovery & prioritization: Run process discovery with supply chain teams (planning, scheduling, inventory, procurement, logistics) to map and size AI opportunities; build a sequenced backlog tied to measurable outcomes. • Hands-on POCs to MVPs: Rapidly build POCs and evolve the best into MVPs using AWS + Databricks + Dataiku + OpenAI, integrating with enterprise data sources and real workflows. • Roadmap ownership: Produce and maintain a capability roadmap (90 days/6 months/12 months) with dependencies, data readiness needs, and adoption/change requirements. • Kinaxis-enabled use cases: Design solutions that complement and enhance Kinaxis workflows (exception management, scenario narratives, decision support, insight generation, planning explainability). • Balance quick wins vs foundations: Deliver short cycle wins that work today while shaping requirements for scalable infrastructure as IT builds out data/AI foundations (reusable components, standard integration patterns, agent orchestration approaches, monitoring/evaluation patterns). • Vision and transition plan: Define a north-star vision and pragmatic transition plan bridging today's constraints to the medium/long-term target state (platform, governance, integration patterns, reusable AI assets). • Production-minded engineering: Define architecture, evaluation metrics, guardrails, monitoring, and last-mile integration patterns; partner with platform/data teams while remaining personally hands-on. • Drive ChatGPT adoption: Build role-based prompt libraries, playbooks, templates, and enablement sessions; measure adoption and iterate based on usage and impact. • Stay current and apply responsibly: Maintain working knowledge of the rapidly evolving AI landscape (GenAI, agents, evaluation, toolchains) and translate relevant advances into practical solutions aligned to enterprise guardrails.

• Map DPD opportunities for AI: Identify and size use cases across drug product operations (execution insights, investigations support, batch record intelligence, documentation acceleration, knowledge retrieval, exception management, production performance). • Build and ship solutions: Deliver POCs/MVPs hands-on - integrating GenAI with operational data and controlled documentation in secure, controlled, and adoption-ready ways. • Roadmap and value cases: Own a DPD AI roadmap with value sizing, dependencies, and adoption plans; align priorities with DPD stakeholders. • Balance quick wins vs foundations: Build what is needed now while shaping and standardizing patterns that will scale with evolving enterprise data/AI infrastructure (reusable components, integration standards, evaluation/monitoring, agent orchestration patterns). • Vision and transition plan: Define the DPD AI north star and a staged plan from POCs to reusable capabilities to scaled solutions as infrastructure matures. • Quality-aware design: Build with regulated realities in mind (traceability expectations, controlled content boundaries, auditability of outputs where required). • Drive ChatGPT adoption: Create role-based prompt kits, playbooks, SOP-adjacent guidance, and enablement sessions; measure adoption and iterate. • Stay current and apply responsibly: Continuously monitor the rapidly advancing AI landscape and selectively bring in relevant capabilities that are safe, practical, and measurable in DPD operations.

• Identify high-leverage DSD use cases: Target opportunities across DS operations (technical knowledge retrieval, investigation support, documentation acceleration, operational insights, exception triage, capacity/coordination decision support). • Build prototypes and deliver MVPs: Create working solutions using AWS/Databricks/Dataiku/OpenAI, integrating operational data and documents into usable workflows. • Data readiness and integration: Assess data availability/quality; define what's needed and influence how to get there (standard sources, access patterns, reusable pipelines). • Roadmap ownership: Produce a DSD capability roadmap with milestones, dependencies, and impact metrics. • Balance quick wins vs foundations: Deliver what helps DSD now while shaping patterns that will scale as the enterprise platform matures (reusable assets, integration standards, evaluation/monitoring, agent orchestration approaches). • Vision and transition plan: Define the DSD AI north star and a staged path from POCs to reusable capabilities to scaled delivery. • Drive ChatGPT adoption: Build prompt libraries, playbooks, and training tied to real DS work; measure usage and outcomes. • Stay current and apply responsibly: Keep pace with AI advances and translate them into practical, controlled, high-impact solutions suitable for DSD environments.