Senior AI Applied Scientist II
$150,000–$177,000 year
HybridVancouver, British Columbia, Canada
Job Summary
Lead end-to-end fine-tuning of large language models on clinical text, radiology reports, and longitudinal patient histories to build specialized domain-adapted models. Design, develop, and operationalize autonomous AI agent systems utilizing multi-step reasoning, tool use, and workflow orchestration to automate complex clinical workflows. Establish robust, automated NLP evaluation frameworks to benchmark clinical accuracy, detect hallucinations, and ensure regulatory compliance of deployed language models. Collaborate on multimodal systems connecting imaging foundation models with LLMs and structured lab data to create holistic patient insight platforms. Own experiment design and validation study scoping, diagnosing failure modes independently while driving data quality standards and clinical annotation criteria. Partner across clinical, annotation, and ML Engineering teams to raise code-quality bars and contribute reusable architectures adopted org-wide. Deliver production-ready LLM and agentic models with documented validation and continuous monitoring benchmarks. This hybrid role is based in Vancouver; no visa sponsorship is provided.
Required Qualifications
- Strong MSc or PhD from a top-tier institution in CS, computational linguistics, AI/ML, statistics, biomedical engineering, or a related field
- A minimum of 4 academic or industry years of ML/NLP experience
- A strong publication or applied-research record including first-author works in top ML/NLP venues (e.g., NeurIPS, ACL, EMNLP, NAACL, ICML, ICLR)
- Expert-level PyTorch
- Proficiency with modern LLM fine-tuning libraries (e.g., Hugging Face Transformers, Unsloth, vLLM, DeepSpeed, Axolotl)
- Hands-on expertise in LLM Fine-Tuning & Alignment: Supervised Fine-Tuning (SFT), Parameter-Efficient Fine-Tuning (LoRA, QLoRA), Reinforcement Learning from Human Feedback (RLHF), and Direct Preference Optimization (DPO) applied to open-source foundation models (e.g., Llama, Qwen, Mistral)
- Demonstrated depth in Agentic AI Architectures: Designing and deploying LLM agents using frameworks like LangGraph, AutoGen, or CrewAI, implementing tool use/calling, ReAct prompting, multi-agent coordination, external memory, and multi-step execution
- Proven track record in NLP Evaluation & Grounding: Designing custom evaluation suites, benchmark creation, automated evaluation metrics (e.g., ROUGE, BERTScore, LLM-as-a-Judge), hallucination mitigation, semantic search, and Retrieval-Augmented Generation (RAG)
- Demonstrated ability to own an LLM/Agent model domain end-to-end (architecting, training, evaluating, deploying to production performance standards)
- Strong collaboration skills across ML Engineering, Product, and Clinical Ops, balancing scientific rigor with rapid delivery
- This hybrid role is based in Vancouver
- Applicants must be legally authorized to work in Canada at the time of hire and must not require employer sponsorship for a work visa (current or future)
Desired Qualifications
- Experience fine-tuning LLMs on medical or clinical text corpora (e.g., EHRs, DICOM structured reports, PubMed)
- Experience integrating vision-language models (VLMs) with medical imaging pipelines
- Familiarity with FDA guidelines for AI/ML-based Software as a Medical Device (SaMD) or clinical decision support tools
- Prior experience in health tech, biotech, or clinical AI environments
Hiring someone like this?
Get your role in front of qualified candidates on Sorce.