← Serch more jobs

Principal Scientist, AI

LinkedIn ChatGPT Jobs San Diego Metropolitan Area
Not Applicable Posted April 17, 2026 2 variants Job link
Thinking about this job
Not Met Priorities
What still needs stronger evidence
Requirements
  • 5+ years delivering ML/AI solutions in life sciences (discovery, translational, clinical, or RWE), including 3+ years leading cross-functional technical teams.
  • Hands-on expertise with Python and core ML/DL frameworks (PyTorch and/or TensorFlow; Keras); strong software engineering practices (testing, code review, version control).
  • Proven experience building production-grade data and deployment pipelines: SQL and Spark, containerization (Docker), orchestration (Airflow/Prefect), cloud services (AWS preferred; Azure/GCP welcome).
  • Experience with multi-agent systems and agent orchestration in production use cases.
  • Track record of rigorous LLM evaluation: designing task-specific benchmarks, implementing automated evaluation frameworks, diagnosing failure modes, and iteratively optimizing retrieval and generation pipelines for accuracy, latency, and cost.
  • Practical GenAI/LLM experience: retrieval-augmented generation, vector databases (e.g., FAISS, Milvus, pgvector), prompt engineering, evaluation frameworks, and safety/guardrail techniques.
  • Strong client-facing skills: translating scientific needs into technical solutions, presenting to senior stakeholders, and contributing to scope and SOWs.
  • Domain fluency with clinical, preclinical, or RWE data and relevant standards (CDISC, OMOP, FHIR) and biomedical ontologies (e.g., OBO, SNOMED, MeSH).
Preferred Skills
  • Experience with knowledge graphs (RDF/OWL, SPARQL, Neo4j) and entity/relationship modeling.
  • Biomedical NLP (e.g., BioBERT, SciBERT) and ontology-driven text mining.
  • Privacy and compliance expertise: de-identification, data use agreements, and audit readiness.
  • Familiarity with data product thinking and monetization of curated datasets.
  • Familiarity with multimodal foundation models in biomedical domains: single-cell embeddings (e.g., scGPT, Geneformer), molecular/chemical LLMs (e.g., ChemBERTa, MolBERT), or medical imaging models (e.g., BiomedCLIP, pathology foundation models).
  • MLOps proficiency with platforms such as AWS SageMaker, Vertex AI, or Kubeflow; experiment tracking (MLflow/Weights & Biases); model registry and monitoring.
Education
  • (Not required) – PhD in Computational Biology, Bioinformatics, Computer Science, Statistics, or related field (or comparable demonstrated relevant experience).