Viva Questions for Computer Science

Computer Science vivas vary significantly depending on whether your research is theoretical, systems-oriented, or applied. Examiners will focus on the rigour of your evaluation, the novelty of your approach, and your ability to position your work against the state of the art. If your work involves building a system, expect detailed questioning about design decisions and benchmarking. If it's more theoretical, be ready to defend your proofs and the assumptions underlying your models.

Computer Science moves fast, and your examiners know that. They'll be interested in how the landscape has shifted during your PhD and whether you've adapted your work accordingly. If a new benchmark, dataset, or competing approach emerged while you were writing up, they'll want to know that you're aware of it and can articulate how it relates to your contribution.

Questions about your research

CS examiners tend to be very specific in their questioning. If you've built a system, they'll want to understand the architecture, the trade-offs, and the evaluation methodology in detail. If you've proposed an algorithm, they'll want to know its complexity, its failure modes, and how it compares with baselines. They'll also ask about reproducibility – could someone else replicate your results from the information in your thesis?

• What is the core research problem you're addressing, and why is it important to the field?
• Can you walk us through your system architecture or algorithmic design, including the key decisions you made?
• How did you evaluate your approach, and why did you choose these specific benchmarks or datasets?
• What baselines did you compare against, and how did you ensure the comparison was fair?
• How does your system or algorithm scale with input size, dimensionality, or complexity?
• What are the key hyperparameters in your model, and how sensitive are your results to them?
• Did you conduct ablation studies, and what did they reveal about which components matter most?
• How did you handle edge cases, adversarial inputs, or failure modes in your system?
• What engineering challenges did you encounter during implementation, and how did you solve them?
• How reproducible are your experiments – could someone replicate them from your thesis alone?
• What computational resources did your experiments require, and how does that affect the accessibility of your approach?
• How did you handle randomness or stochasticity in your experiments – did you report variance across runs?

Questions about theory and literature

Computer Science examiners will expect you to know the state of the art in your area thoroughly. They'll ask how your approach differs from closely related work, whether there are theoretical guarantees for your method, and what assumptions you've made. If your area has moved quickly during your PhD – as is common in fields like machine learning or systems – they'll want to see that you've engaged with recent developments, even if they appeared after your main experiments were complete.

• How does your work relate to the current state of the art, and what specifically does it improve?
• What are the most closely related approaches in the literature, and where exactly does yours differ?
• Are there theoretical guarantees for your approach – bounds on performance, convergence, or correctness?
• What assumptions does your model or system make, and how realistic are they in practice?
• How has the field moved during your PhD, and has that affected the framing or significance of your contribution?
• Are there techniques from adjacent fields – statistics, operations research, control theory – that could have been applied here?
• How does your work engage with recent developments that emerged during your write-up period?

Questions about contribution and impact

In Computer Science, contribution can take many forms – a new algorithm, a new system, a new dataset, a new theoretical result, or a new way of framing a problem. Examiners will want you to be precise about what is genuinely novel. They'll also be interested in the practical impact of your work – whether it could be deployed in industry, open-sourced, or used by other researchers.

• What is your thesis's primary contribution to the field, and how would you distinguish it from incremental improvement?
• Could your approach be applied to other domains, problem types, or scales?
• What are the practical implications of your work for industry, open-source communities, or other researchers?
• If you were to distil your thesis into a single conference paper, what would the key message be?
• How does your work open up new research directions that didn't exist before?
• Have you released code, data, or tools that others can build on?

Tough follow-ups your examiners might ask

CS examiners are good at finding the gap between what your evaluation shows and what you claim. They'll push on whether your results generalise beyond the benchmarks you used, whether a simpler approach might work just as well, and whether your method would survive contact with real-world data. Be prepared to discuss the limitations of your evaluation honestly.

• Your evaluation shows improvement on this benchmark – but would it generalise to real-world, noisy data?
• What happens when the assumptions your model relies on are violated in practice?
• A much simpler baseline might achieve comparable results – how do you justify the added complexity of your approach?
• How would your system behave under adversarial conditions or distributional shift?
• What would you do differently if you were starting this project today, knowing what you know now?
• How do you respond to the argument that your results are benchmark-specific rather than genuinely general?
• If a major new dataset or benchmark emerged tomorrow, how confident are you that your approach would still perform well?

Ready to practise? These are the kinds of questions your examiners will ask – but in a real viva, they won't stop at the first answer. They'll follow up, probe deeper, and test how well you can think on your feet. Try VivaCoach to practise with AI-powered follow-up questions tailored to your thesis.

Practise with AI-powered follow-up questions tailored to your thesis.