About XpertDox
XpertDox, Inc. is a physician-founded, venture-backed healthcare AI company headquartered in Scottsdale, Arizona, with a global delivery team in India. Our platform, XpertCoding, autonomously codes outpatient medical claims directly from clinical documentation using neural networks, machine learning, and natural language processing.
Position Summary
You will work on neural network models that read clinical documentation and predict medical codes. Your first job is diagnosis: finding out why these models get specific cases wrong, and testing fixes until one holds up. Once a fix is proven, you write it into production code that runs inside the automated coding pipeline. This is an individual contributor role with no team management. It suits someone who wants to understand model errors and fix them. It does not suit someone who wants to run experiments without shipping, or who works only in notebooks.
Key Responsibilities
In this role you will:
• Own diagnosis of model errors on clinical text: find where predictions go wrong and trace each pattern to its root cause.




• Design experiments that test one specific fix at a time, and decide from the results whether the fix moves forward.
• Train, fine-tune, and evaluate neural network models for multilabel classification on clinical documentation.
• Improve precision on imbalanced classes, where a wrong code has a real cost.
• Write production Python that moves proven fixes into the automated coding pipeline.
• Explain the tradeoffs of a proposed change to the AI Team Lead before it touches a production model.
• Document every experiment: what you tried, what happened, and what you decided.

Required Qualifications
• At least 1 year of hands-on machine learning work outside coursework.
• Hands-on experience fine-tuning pretrained language models for text classification.
• Project work training deep learning models in Python.
• Experience with multilabel classification, including handling class imbalance.
• Pr

📌 AI Data Scientist (Noida)
🏢 XpertDox
📍 Noida