How to Clear Your Final Year Viva Without Panicking
The 30 questions every evaluator asks — and how to answer them confidently. Tested across 7,000+ vivas at Anna University-affiliated colleges.
The truth about vivas: Evaluators are not trying to fail you. They are checking that you understand what you submitted. If you know your project well enough to explain every component, protocol, and result, you will pass — even if the demo glitches.
6 Rules Before Your Viva
- Know your block diagram cold — draw it on the whiteboard if asked.
- Understand every component in your circuit. "I don't know what this does" is an instant mark deduction.
- Prepare a 2-minute project summary. Practice it until it sounds natural.
- Do at least one mock viva with a friend who has no idea what your project does.
- Bring your test results — actual numbers, not "it worked fine."
- If the demo fails: stay calm, explain what it does when working, and describe the likely failure mode professionally.
About Your Project
Q: Explain your project in one sentence.
A: State: what it does, what it uses, and who benefits. Example: "This system detects defects in PCBs in real time using a MobileNet model deployed on Raspberry Pi 4, eliminating manual quality inspection."
Q: What is the objective of your project?
A: List 3–4 specific, measurable objectives. Not "to build an IoT system" but "to monitor soil NPK levels and automate irrigation with <5% false trigger rate."
Q: What problem does your project solve?
A: Frame it as: current problem + your solution + impact. "Current glucose monitoring is invasive — our NIR-based system eliminates needle pricks, improving patient compliance."
Q: What is the scope of your project?
A: Scope = what is included + what is not. "The scope includes real-time sensor data acquisition, edge inference, and cloud logging. It does not include mobile app development."
Technical Questions
Q: Why did you choose this microcontroller / processor?
A: Compare it to one alternative. "We chose STM32F4 over Arduino Mega because it has a hardware FPU for signal processing, 168 MHz clock speed, and multiple SPI/I2C peripherals natively."
Q: What is the communication protocol you used and why?
A: Name the protocol, explain its advantages in context. "We used MQTT over HTTP because MQTT is lightweight (2-byte header), supports publish-subscribe for multiple subscribers, and works well on low-bandwidth networks."
Q: What is the accuracy of your model / system?
A: Give a number — then explain how you measured it. "Our model achieved 94.3% accuracy on a 500-image test set. We used an 80/10/10 train/val/test split."
Q: How did you test the project?
A: Mention unit testing, integration testing, and the conditions under which you tested. "We tested each sensor module individually, then ran full-system tests under 3 lighting conditions and 5 temperature settings."
IEEE Paper Questions
Q: What is the base paper for your project?
A: State the title, authors, journal, and year. "Our base paper is 'Edge-AI Based Defect Detection System Using MobileNetV2' by Zhang et al., IEEE Access, 2023."
Q: How is your implementation different from the base paper?
A: Your differentiation is crucial. "The paper used a GPU server; we deployed the quantized model on Raspberry Pi 4 using TFLite — achieving 15 FPS on-device with no cloud dependency."
Q: What are the limitations of your project?
A: Be honest and specific. "The model struggles with specular reflections on metallic surfaces and requires controlled lighting. A future improvement would include polarized illumination."
Q: What future work can be done on this?
A: Suggest 2–3 concrete improvements. "Future work includes adding a 6-DOF robotic arm for auto-reject, expanding the dataset to 5,000+ images, and testing on NVIDIA Jetson for faster inference."
Panic Questions
Q: Did you build this yourself?
A: "Yes — our team worked on this project from topic selection to final testing. I handled [your specific part], and I can walk you through any part of the implementation."
Q: Why is this output wrong / inconsistent?
A: Never say "I don't know." Say: "That could be due to [sensor noise / lighting variation / model drift]. In our tests, we observed this under [specific condition] and mitigated it by [approach]."
Q: What is the cost of your project?
A: Know your BOM. Break it down: "Hardware: ₹4,200 (sensors ₹1,800, MCU ₹1,200, PCB ₹1,200). Software: open source. Total: ₹4,200."
Q: Can this be commercialized?
A: "Yes — with minor modifications. The target market is [manufacturing / agriculture / healthcare]. We estimated a unit cost of ₹X at 1,000-unit production scale, making it viable for SMEs."
We Prepare You for Every Question
Our Standard and Premium packages include 30+ viva Q&A specific to your project and a mock viva session before your actual date.
Get Viva Preparation Support