LLM Trainer for Medical Data Extraction (MIMIC-4)

About OpenTrain

OpenTrain is the #1 platform for building careers in AI training and data labeling. We connect contributors with hands-on projects that teach and refine state-of-the-art AI systems while offering flexible, remote work that fits around your life.

For this engagement, OpenTrain AI is the hiring organization and will manage onboarding, instructions, and payment.

Why AI training in medicine matters

AI training (data annotation and human-in-the-loop work) is the human foundation of modern medical AI: models learn from curated clinical examples and human feedback. Medical projects let you work on real clinical data, help shape model behavior that could influence care-oriented tools, and build rare expertise at the intersection of healthcare and machine learning.

This role focuses on extracting structured clinical values to train a model that predicts the occurrence of a clinical event—work that directly supports safer, more reliable medical AI systems.

The role

OpenTrain AI is hiring a contractor to collect specific clinical data points from the MIMIC-4 database and prepare them for LLM training. You will use MedPALM2 or BioBERT as part of the extraction/training workflow to assemble a dataset of urine output and serum creatinine values tied to patient events.

This is a fixed-price contract (USD 1,000). The project is open to applicants worldwide. The engagement level is intermediate — you should be comfortable working with clinical data and model-training tools, but prior specialized experience is preferred, not required.

• Employment type: Contractor
• Data type listed for this task: MEDICAL_DICOM
• Labeling task type: DATA_COLLECTION
• Worldwide applicants accepted

What you'll do

Follow provided protocols and tooling to locate, extract, and validate urine output and serum creatinine data points from MIMIC-4. Prepare the data in the required format for downstream LLM training and quality review.

You may use MedPALM2 or BioBERT components as part of the extraction and labeling workflow; follow the project's technical guidance and document any edge cases, ambiguities, or data quality issues you encounter.

• Extract urine output and serum creatinine values from MIMIC-4 according to provided specs
• Map and format records for LLM training input (timestamps, units, patient linkage as specified)
• Flag missing, inconsistent, or implausible values and document resolution steps
• Work with the project lead to confirm dataset completeness and submission

Requirements

You must be comfortable working with clinical datasets and follow data-handling and privacy instructions provided by OpenTrain AI. The role is intermediate-level: prior hands-on experience with medical LLM or NLP model training (e.g., MedPALM2, BioBERT, clinical NLP) is preferred but not mandatory.

Because this work involves clinical data, attention to detail, careful documentation, and the ability to follow technical extraction protocols are essential.

• Intermediate experience level with data extraction or model-training workflows
• Prior experience with medical LLM/NLP projects preferred but optional
• Strong attention to data quality, units, and timestamp consistency
• Willingness to follow secure data-handling instructions and project protocols

Who should apply

This project is a good fit for data engineers, clinical informaticists, NLP practitioners, and experienced annotators who have worked with clinical datasets or healthcare-focused models.

If you have hands-on experience with MIMIC datasets, clinical variable curation, or with tools like MedPALM2/BioBERT, you’ll be especially well suited. However, applicants with solid data extraction skills and a careful, methodical approach are encouraged to apply even without prior medical-LLM work.

How the project works & payment

On acceptance you will receive detailed instructions, extraction specifications, and any tooling access required. Deliverables will be reviewed for completeness and data quality before final payment.

This is a fixed-price engagement: total payment is USD 1,000. OpenTrain AI will manage contracting and payment after submission and quality approval. No additional compensation or guarantees are implied beyond the fixed price.

• Fixed-price contract: USD 1,000 total
• Deliverables and quality checks required for final payment
• Open worldwide — contractor status