Clinical AI Engineering Intern

Lumeris

5d•$54,800 - $73,250•Hybrid

About The Position

Lumeris is physician-founded, mission-driven, and building the future of primary care. Our platform, Tom, is an agentic AI that works alongside care teams — automating routine tasks, surfacing the Best Next Action for every patient in real time, and delivering proactive outreach before, during, and after patient visits. Built on billions of clinical data points and over 100 EHR integrations, Tom powers Primary Care as a Service — helping health systems improve outcomes, reduce burden on physicians, and deliver value-based care at scale. As an AI/ML Intern at Lumeris, you will work directly on the technical core of Tom — building and fine-tuning large language models using real-world longitudinal clinical data. The goal: expand Tom’s predictive capabilities, with specific targets including 30-day readmission rates and time-to-event modeling. This is not a passive learning experience. You will be an active contributor to a defined, high-impact research initiative running on Google Cloud Platform, working alongside ML engineers, data scientists, and clinical informatics experts. The work is experimental, scientifically rigorous, and directly tied to patient outcomes. Applicants must be based in the greater Boston area with ability to travel to Cambridge 3-days per week.

Requirements

Currently pursuing a Bachelor’s or Master’s degree in Computer Science, Data Science, Computational Biology, Biomedical Informatics, or a closely related field.
Coursework or hands-on experience in machine learning, deep learning, or neural networks.
Foundational understanding of how LLMs work, including transformer architectures and model training pipelines.
Strong analytical and problem-solving skills with a scientific, experiment-driven mindset.
Effective communication skills and ability to work collaboratively in a research-oriented team environment.

Nice To Haves

Experience with LLM fine-tuning techniques (e.g., LoRA, PEFT, instruction tuning, supervised fine-tuning).
Familiarity with Google Cloud Platform (GCP) or other cloud environments for ML workloads.
Experience with Survival Modeling Techniques
Experience with standard medical terminologies like OMOP, FHIR
Background in clinical informatics, bioinformatics, or healthcare data.
Proficiency in Python and ML frameworks such as PyTorch, TensorFlow, or HuggingFace Transformers.
Prior research experience (academic, industry, or self-directed) — particularly in applied ML or healthcare AI.
Exposure to embedding models, vector databases, or sequence-based modeling.
Graduate-level coursework in NLP, deep learning, or biomedical data science is a strong plus.

Responsibilities

Fine-tune and train large language models using longitudinal clinical patient data on GCP infrastructure, with a focus on expanding Tom’s predictive capabilities.
Contribute to the design and experimentation of transformer-based architectures and embedding models aimed at predicting specific clinical outcomes.
Assist in evaluating model performance against defined prediction targets including 30-day readmissions and time-to-event events.
Operate in an experiment-driven development environment — designing tests, iterating on model behavior, tracking results, and documenting findings with scientific rigor.
Explore and apply LLM fine-tuning techniques (e.g., LoRA, PEFT, supervised fine-tuning) appropriate to the clinical data and prediction objectives.
Contribute to the team’s working knowledge of embedding models, sequence-based architectures, and clinical data preprocessing.
Work cross-functionally with ML engineers, clinical informatics specialists, and data scientists in a highly collaborative research environment.
Engage with a structured mentorship model — supported by project mentors and a dedicated technical mentor throughout the engagement.
Prepare and present a formal project deliverable at the conclusion of the internship, summarizing methodology, outcomes, and recommendations.