About the positionThe AI Engineer at Abbott will accelerate proof-of-concepts (PoCs) across Diabetes Care products and internal enterprise solutions. Our focus is applying Generative AI, AI agents, and Machine Learning to improve experiences, decision‐making, and efficiency—both in customer/product contexts and in internal processes (e.g., documentation, quality workflows, analytics, operational automation).
This role is AI‐first : you're expected to use AI tools in your daily work to speed up delivery while maintaining engineering rigor, traceability, and quality.
Responsibilities
Build end-to-end AI workflows : data → model/agent logic → evaluation → deployable prototype.
Develop AI agents that use tools (function calling, retrieval, routing, multi‐step plans, state/memory, workflow orchestration).
Apply AI first principles : model behavior, limitations, grounding strategies, uncertainty handling, prompt injection awareness, and safe‐by‐design patterns.
Design and run evaluations : golden datasets, automated checks, prompt/agent regression tests, and human‐in‐the‐loop review when needed.
Implement fine‐tuning / adaptation workflows when appropriate (dataset prep, training runs via managed services, versioning, validation).
Build and compare ML approaches (baselines, feature pipelines, metrics, error analysis) and combine them with GenAI when useful.
Integrate PoCs into real systems via APIs/services, and instrument for monitoring (latency, cost, quality).
Produce clear demos and documentation so results translate into go/no‐go decisions and scalable next steps.
Requirements
Strong Python engineering : clean code, debugging, testing discipline, ability to ship prototypes quickly.
Hands‐on GenAI/LLM experience using cloud APIs and delivering solutions beyond notebooks.
Proven experience building AI workflows and agents that use tools (orchestration, routing, structured outputs, state handling).
Strong understanding of AI first principles (why models fail, hallucinations, grounding, tradeoffs, evaluation‐driven development).
Experience with evaluation and testing for AI systems (unit/integration tests + model‐quality evaluation).
Experience with fine‐tuning or model adaptation workflows (and knowing when not to fine‐tune).
Solid machine learning fundamentals (data prep, training/inference, metrics, baseline comparisons, model selection).
Strong communication skills: can explain results, risks, and tradeoffs to technical and non‐technical stakeholders.
#J-18808-Ljbffr