AI Architect - Modern AI Applications

• →

  Architect, design and develop AI platform to support large language models to handle healthcare-specific language, regulatory requirements, and ethical considerations.

• →

  Frameworks which can support AI pipelines that can process structured and unstructured healthcare data (FHIR, HL7, clinical notes, claims data).

• →

  Contribute to domain-specific model architectures that improve clinical decision-making, revenue cycle management, and patient engagement.

• Serve as the primary technical authority on any AI Platform related development within product teams.

• Mentor junior AI designers and engineers through code reviews, research guidance, and technical workshops.

• Drive internal knowledge-sharing on emerging AI trends, frameworks, and best practices.

• Implement rigorous model evaluation frameworks for accuracy, robustness, and fairness.

• Ensure compliance with healthcare privacy and data security regulations (HIPAA, HITRUST).

• Partner with engineering to move research prototypes into production environments.

• Master’s in Computer Science, AI, Machine Learning, or related field AND 10+ years of experience in developing AI/ML platform on Cloud & on-Prem.

Or

• 15+  software engineering background in  AI, Machine Learning, or related field with 10+ years of experience in building AI/ML platforms.

• Proficiency in Python, and modern ML libraries.

• Experience with GenAI, LLMs, transformer architectures, and advanced Model Routing (dynamically selecting and orchestrating open-source vs. proprietary models based on latency, cost, and capability tiers).

• Skilled with AI development tools and frameworks (PyTorch, TensorFlow, Hugging Face, LangChain, LlamaIndex).

• Deep expertise in building enterprise-grade RAG systems, including advanced chunking, hybrid search, vector database management, and retrieval optimization.

• Hands-on experience with autonomous AI agents and reasoning systems, specifically mastering Agent-to-Agent (A2A) communication protocols and the Model Context Protocol (MCP) for multi-agent orchestration.

• Advanced skills in Agent Context Management, including context window optimization, stateful memory injection, caching strategies, and managing long-running agent contexts.

• Proven ability in Observability, Logging, and Scalable systems, including specialized LLM/Agent tracing to monitor reasoning steps, token usage, and system latency.

• Strong track record applying AI architectures to scalable, generic enterprise platforms and complex use cases.

• Strong background with cloud platforms, On-Prem deployments, and MLOps/LLMOps practices.

• Strong background in containerization and infrastructure as code (Kubernetes, Terraform, etc.).

• Experience with Knowledge Graph DBs (e.g., Neo4j), NoSQL/SQL databases, and native Vector databases, particularly in designing advanced GraphRAG and hybrid-retrieval architectures.

• Familiarity with advanced multi-agent orchestration frameworks (e.g., LangGraph, AutoGen, CrewAI, Semantic Kernel) for complex, stateful workflow execution.

• Proven ability to take complex AI Architecture/Design—especially non-deterministic LLM and multi-agent systems—from concept to highly available, production-grade deployments.

• Experience with LLM Evaluation frameworks (e.g., Ragas, TruLens, LLM-as-a-Judge) to continuously monitor and score agent reasoning and RAG retrieval quality.

• Knowledge of model fine-tuning techniques (LoRA, PEFT) and model distillation to create smaller, task-specific models that optimize cost and latency within the model routing layer.

• Familiarity with implementing platform-wide AI guardrails, prompt injection defenses, and output validation mechanisms.

Carefully read these Terms of Use before using this website. Your access to and use of this website and application for a job at Solventum are conditioned on your acceptance and compliance with these terms.

Please access the linked document by clicking here. Before submitting your application you will be asked to confirm your agreement with the
terms.