1. What is an AI Engineer at The Johns Hopkins University?

As an AI Engineer at The Johns Hopkins University, specifically within the Department of Radiology-Diagnostic Imaging Division, you are at the forefront of translating cutting-edge artificial intelligence into tangible clinical outcomes. This role, officially titled Sr. Radiomics and AI Engineer, is not just about writing code; it is about building the bridge between complex computational models and life-saving medical insights. You will be directly responsible for developing algorithms that analyze medical images to extract clinically relevant features, empowering radiologists and oncologists to make faster, more accurate diagnoses.

Your work will have a profound impact on patient care and medical research. Operating primarily out of the Felix lab, you will handle high-dimensional clinical data, build robust data preprocessing pipelines, and deploy deep learning models that function reliably in real-world healthcare scenarios. The scale and complexity of this role are immense, as medical imaging data requires rigorous validation, strict adherence to privacy standards, and models that generalize across diverse patient populations.

What makes this position uniquely compelling is its highly multidisciplinary nature. You will not be siloed in an engineering department. Instead, you will work shoulder-to-shoulder with world-renowned healthcare professionals, translating their clinical needs into technical architectures. If you are passionate about using machine learning to push the boundaries of medical science and thrive in an environment that demands both academic rigor and engineering excellence, this role at The Johns Hopkins University is an exceptional opportunity.

2. Common Interview Questions

3. Getting Ready for Your Interviews

Preparing for an interview at The Johns Hopkins University requires a balanced focus on advanced machine learning concepts, software engineering fundamentals, and domain-specific knowledge in healthcare. You should approach your preparation by mastering the following key evaluation criteria:

Technical and Domain Expertise Interviewers will rigorously assess your proficiency in machine learning and deep learning, particularly as applied to computer vision and radiomics. You must demonstrate a deep understanding of how to handle medical imaging formats, extract features, and build models using Python, C++, or MATLAB. Strong candidates will effortlessly discuss the nuances of 3D image processing and the mathematical foundations of their chosen algorithms.

Systems and Pipeline Engineering Because you will be maintaining the computational and storage resources in the Felix lab, your ability to build and manage robust pipelines is critical. Interviewers evaluate your familiarity with software development best practices, including version control, testing, and cluster management. You can showcase strength here by discussing how you design scalable, reproducible data preprocessing pipelines that ensure high-quality input for clinical AI models.

Cross-Functional Collaboration Working in a medical research environment means you must translate technical jargon into clinical reality. You will be evaluated on your ability to communicate complex AI concepts to non-technical stakeholders, such as radiologists and oncologists. Demonstrate this by sharing examples of how you have collaborated across disciplines to define project requirements or troubleshoot unexpected model behaviors in real-world scenarios.

Scientific Rigor and Validation In healthcare, a model's failure can have serious implications. You will be judged on your approach to testing and validating algorithms using clinical datasets. Interviewers look for candidates who prioritize model interpretability, robustness, and continuous performance monitoring over simply achieving high accuracy on a training set.

4. Interview Process Overview

The interview process for the Sr. Radiomics and AI Engineer role at The Johns Hopkins University is thorough and highly collaborative, reflecting the multidisciplinary nature of the position. You will typically begin with a recruiter phone screen to verify your background, minimum qualifications, and alignment with the salary expectations and onsite requirements in Baltimore, MD. This is followed by a technical screen, often conducted via video call, where you will discuss your past projects, deep learning fundamentals, and approach to medical imaging challenges.

If you progress to the onsite or virtual panel rounds, expect a rigorous series of interviews that blend technical deep dives with behavioral and clinical collaboration assessments. A hallmark of the JHU process is the research or project presentation. You will likely be asked to present a past project to a mixed panel of engineers and clinicians, defending your architectural choices and explaining the clinical relevance of your work. The process is designed to test not just your coding ability, but your capacity to thrive in a high-stakes, cross-functional medical research environment.

This visual timeline outlines the typical progression from initial screening to the final panel interviews. You should use this to pace your preparation, focusing first on core ML concepts for the technical screen, and then shifting to presentation skills and domain-specific medical imaging knowledge for the final rounds. Note that because you will be working closely with the Felix lab and clinical staff, the final rounds will heavily emphasize your communication skills and cultural fit within a medical institution.

5. Deep Dive into Evaluation Areas

To succeed, you must deeply understand the specific technical and behavioral areas your interviewers will probe. The evaluation is tailored to the unique demands of applying AI to radiology.

Machine Learning and Computer Vision for Healthcare

This is the core of your technical evaluation. Interviewers want to see that you can build, train, and deploy deep learning models specifically optimized for imaging data. Strong performance means you can discuss the architecture of Convolutional Neural Networks (CNNs), segmentation models like U-Net, and how to handle the severe class imbalances often found in medical datasets. Be ready to go over:

• Image Segmentation and Classification – Techniques for isolating tumors, organs, or anomalies in medical scans.
• Handling High-Dimensional Data – Strategies for processing large 3D and 4D datasets efficiently.
• Model Interpretability – Methods like Grad-CAM or saliency maps to explain model predictions to clinicians.
• Advanced concepts (less common) – Generative Adversarial Networks (GANs) for synthetic medical data generation, self-supervised learning for unannotated clinical data. Example questions or scenarios:
• "Walk me through how you would design a deep learning architecture to segment lung nodules from 3D CT scans."
• "Medical datasets often have massive class imbalances. How do you adjust your training pipeline and loss functions to account for this?"
• "Explain how you would validate a model to ensure it doesn't overfit to the specific imaging equipment used in our hospital."

Radiomics and Domain Knowledge

You are not just an AI engineer; you are a Radiomics Engineer. You must understand how to extract quantitative features from medical images and correlate them with clinical outcomes. Interviewers will look for your familiarity with medical data formats and your ability to speak the language of radiology. Be ready to go over:

• Medical Imaging Formats – Working with DICOM, NIfTI, and understanding metadata.
• Feature Extraction – Traditional radiomic feature extraction (shape, texture, intensity) versus deep learning feature extraction.
• Clinical Workflow Alignment – Understanding how AI tools integrate into a radiologist's PACS (Picture Archiving and Communication System). Example questions or scenarios:
• "How do you handle artifacts or varying resolutions across different MRI sequences before feeding the data into your model?"
• "Describe a time you used traditional radiomic features alongside deep learning embeddings. Why did you choose that approach?"

Software Engineering and Infrastructure

Because you will maintain the cluster, computational, and storage resources in the Felix lab, your engineering fundamentals must be rock-solid. Interviewers will evaluate your ability to write clean, production-ready code and manage infrastructure. Be ready to go over:

• Programming Proficiency – Writing optimized code in Python, C++, or MATLAB.
• Data Pipelines – Building automated, reproducible preprocessing pipelines.
• Cluster Management – Experience with Linux environments, job scheduling (e.g., SLURM), and containerization (Docker). Example questions or scenarios:
• "How do you structure your version control and testing for an AI model before deploying it for clinical validation?"
• "Imagine our GPU cluster in the Felix lab is running out of storage due to massive imaging datasets. How do you architect a more sustainable data management strategy?"

Cross-Disciplinary Collaboration

The success of your algorithms depends on your ability to work with radiologists and oncologists. Interviewers will test your empathy, patience, and ability to translate technical limitations into clinical risks. Be ready to go over:

• Requirement Gathering – How you elicit features and constraints from medical professionals.
• Communicating Results – Explaining false positives, false negatives, and confidence intervals to non-engineers.
• Navigating Ambiguity – Managing projects where clinical definitions of a disease may be subjective or evolving. Example questions or scenarios:
• "Tell me about a time you had to explain a complex machine learning concept to a stakeholder with no technical background."
• "If a radiologist disagrees with your model's output, how do you handle the situation and investigate the discrepancy?"