Advancing Medicine with
Machine Learning

Tensor Lab

We pair machine learning students with medical students to tackle real clinical problems, guided by internationally-renowned faculty mentors. Join us to lead an independent research project and have the opportunity to publish your discoveries.

01 // THE APPROACH

How It Works

Projects are organized into triads: a ML student who builds, a medical student who guides, and a physician who ensures it's clinically sound.

You Build It

Machine Learning Fellow

You write the code and run the experiments, guided by a team of experienced advisors and surrounded by a cohort of equally-talented peers.

Join as Engineer
Clinical Context

Medical Student

You bring the research question, explain how doctors actually think, and help frame results for a clinical audience.

Join as Medical Student
Quality Control

Faculty Physician

As a practicing physician, you review methods, ensure the research is clinically valid, and open doors for publication.

Partner With Us
02 // TIMELINE

Timeline

Dec 2025 – Jan 2026

Medical Students Submit Projects

If you're a medical student with a research idea, submit a proposal describing the clinical problem you want to solve.

Feb 2026

Machine Learning Students Apply

Browse projects and apply to the ones you're interested in. Medical students interview applicants and select their fellow.

Mar 2026

Teams Matched

Once matched, you'll meet your team and start planning the project scope and timeline.

Apr – May 2026

Institutional Onboarding

Members of your team will get neccessary credentials and access to the tools they need to get started.

Jun – Aug 2026

10-Week Research Sprint

Build the model, run experiments, analyze results, and write the paper. This is the core of the fellowship.

Sep 2026

Present your work

Share your findings at the Tensor Lab Symposium and submit your paper for publication

03 // OUR TEAM

Leadership Board

Founding Directors

Matt Allen Matt Allen Co-Founder & Executive Director University of California, San Francisco
School of Medicine
Aaron Ge Aaron Ge Co-Founder & Technical Director University of Maryland
School of Medicine

Executive Board

Chy Murali Chy Murali Operational Director University of Maryland
School of Medicine
Gavin Shu Gavin Shu Strategic Director University of California, San Francisco
School of Medicine
Angie Lee Angie Lee Business Director University of Maryland
School of Medicine
04 // 2025 COHORT

Selected Works

From our Pilot Cohort
Benchmarking / ICU Data NCI

LLMs vs. Traditional ML

Fellow: Jialong • Mentor: Aaron Ge

Compared the performance of LLM-based semantic models (using LLM embeddings) against traditional ML models (XGBoost) across different clinical prediction tasks.

Result: Text representation critically affects performance; traditional ML excels at physiological data.
Pharmacogenomics UCSF

Oncology Knowledge Graph

Fellow: Dmitry • Mentor: Diem My

Developed the OPMG prototype to accelerate clinical evidence synthesis using an ETL-First strategy and Knowledge Graph RAG system.

Result: Validated ETL-first strategy and end-to-end KG-RAG functionality.
Oncology / SDOH UCSF

Prostate Cancer & SDOH

Fellow: Malavika • Mentor: Chy Murali

Trained ML models using SEER registry data linked with Social Determinants of Health proxies to stratify prostate cancer patients.

Result: SDOH adds measurable predictive value, highlighting social context.
Head & Neck Cancer UMSOM

RFS Prediction via PET/CT

Fellow: Koushik • Mentor: Laura Chen

Investigated whether a simplified multimodal model using only PET/CT and clinical data could achieve comparable RFS prediction to expert-annotated models.

Result: Simpler PET/CT model achieved comparable performance and is more generalizable.
Emergency Medicine Kaiser

LLM ER Triage Simulator

Fellow: Ashwin • Mentor: Aaron Ge

Developed a training simulator using LLMs to create interactive, data-grounded emergency triage encounters from MIMIC-IV data.

Result: Confirmed feasibility of using LLMs for realistic medical training tools.
Cardiology / Readmission UCSF

Predicting Readmissions

Fellow: Shravya • Mentor: Griffith Hughes

Developed an attention-based deep learning model (BioClinicalBERT) to predict 30-day readmissions from free-text discharge summaries.

Result: Achieved AUROC 0.730 with interpretable predictions via SHAP.
Lung Cancer UCSF

Deep Learning for CT Lung

Fellow: Juan Arturo • Mentor: Matt Kim

Compared ResNet50, VGG16, and InceptionV3 for CT-based lung lesion classification using the IQ-OTH/NCCD dataset.

Result: InceptionV3 showed the most consistent performance (0.92 F1).

Fellow Stories

“

Tensor Lab showed me meaningful challenges that combine both clinical and technical domains. As I prepare for my PhD applications, this experience was invaluable in clearly defining my research direction.

Jialong Research Fellow
“

Tensor Lab taught me to conceptualize and develop applications for LLMs that require significant domain knowledge.

Ashwin Research Fellow
“

The guidance from the faculty and mentors was exceptional... I wouldn’t have reached our final publication goals without their help.

Koushik Research Fellow
“

Collaborating with a global cohort... provided me with a network of contacts for future projects.

Juan Arturo Research Fellow
05 // FAQ

Common Questions

Do medical students need to know how to code?

No. As a medical student, you bring the research question and clinical expertise. We match you with a Machine Learning fellow who handles all the coding.

How much time do medical students spend per week?

About 2-4 hours per week. You'll meet with your fellow regularly to review progress and guide the direction of the research.

Do Machine Learning fellows need a medical background?

No. Your medical student partner will teach you everything you need to know about the clinical context. We're looking for strong Python/ML skills.

How much time do Machine Learning fellows spend per week?

Plan for 20 hours per week during the 10-week summer sprint. Think of it like a serious research internship.

Is this remote or in-person?

Mostly remote. You'll meet with your team over Zoom/Discord, but some chapters organize optional in-person sessions.

Is this paid?

Not yet. Right now, this is an unpaid fellowship focused on research experience and publication. We're working on funding for future cohorts.

Who owns the code and research?

We believe in open science. Code is open-sourced, and all team members share authorship. Patient data stays under the PI's institutional controls.

06 // OPEN ROLES

Join the Team

For Machine Learning Students: Join Waitlist

RESEARCH FELLOWSHIP • PUBLICATION OPPORTUNITY

Applications open February 2026. Join the waitlist to get notified and browse cutting-edge clinical AI projects. You'll work with actual patient data, collaborate with medical students who guide the clinical direction, and publish research under renowned faculty mentors at top institutions. We provide the clinical expertise—you bring the technical skills to build solutions that matter.

What You'll Do

  • Build production-grade ML models with real medical datasets
  • Collaborate with medical students who guide the clinical direction
  • Work under renowned faculty mentors at top institutions
  • Co-author and submit papers for peer-reviewed publication

Who Should Apply

  • Students with strong Python/ML experience (PyTorch, scikit-learn, TensorFlow)
  • Students committed to 15 hrs/week during the 10-week summer program
  • Students excited to apply ML to high-impact healthcare challenges

For Medical Students: Lead a Research Project

MENTORSHIP ROLE • CO-FIRST AUTHORSHIP

Applications open now. Turn your clinical insights into published research by leading an AI-powered project. We'll pair you with a talented Machine Learning fellow who builds the technical solution while you guide the clinical direction. If you don't have a faculty mentor yet, we'll match you with one. You bring the medical knowledge—they bring the machine learning expertise. Together, you'll co-author a paper that advances both fields.

What You'll Do

  • Define a meaningful clinical problem that Machine Learning can address
  • Interview and select a Machine Learning fellow to collaborate with
  • Guide research direction through weekly strategy meetings
  • Co-write the manuscript and submit for peer-reviewed publication

Who Should Apply

  • First and second-year medical students with dedicated research time (summer block or gap year)
  • Students passionate about applying AI to solve real clinical challenges
  • Students ready to earn co-first authorship on an impactful publication

Start a Chapter at Your Medical School

Want to bring the Tensor Lab to your institution? We're looking for teams of 3+ students to launch new chapters. Below are the available chapter directorship positions.

Executive Director

Strategy & Relationships
  • Works with medical school administration to get institutional approval
  • Recruits faculty mentors and medical students
  • Manages the chapter and handles any issues
We'd love to hear from you if: You're organized, entrepreneurial, and good at building relationships.

Technical Director

Quality Assurance
  • Reviews project proposals to make sure they're technically feasible
  • Evaluates Machine Learning applicants (look at their GitHub, past projects)
  • Sets up code standards and templates
We'd love to hear from you if: You have a CS/ML background and enjoy helping others debug.

Operations Director

Logistics & Workflow
  • Manages the application portal and deadlines
  • Routes applications to mentors for review
  • Sends acceptance letters and onboarding materials
We'd love to hear from you if: You're detail-oriented and like building systems that work.
Contact Us to Learn More