Spandan
Live lecture speech turned into real-time class polls, with engagement analytics.
Ph.D. Scholar · IIT Ropar
I study what live classrooms reveal about learning, and build the tools that capture it.
Computing education research, live classroom polling and learning analytics, with the systems to match.
Research focus
Every live poll leaves a record. My work reads that record carefully, to see what questions do in a lecture, how good they are as measurements, and what learners reveal beyond right or wrong.
Interpreting each live poll against the lecture it came from, to see what it actually does in the class.
Treating classroom questions as a measurement instrument: how reliable they are, and where answer bias creeps in.
Using response timing and how a class divides to understand learners, even before an answer key exists.
Featured work
Live lecture speech turned into real-time class polls, with engagement analytics.
Question-driven teaching: student questions steer the class, with teacher insights.
Anonymous peer grading with anomaly detection and TA-to-teacher escalation.
The VLED Lab's public site: products, research, events and team.
Publications
arXiv:2609.38907 · Sep 2026 · cs.HC, cs.AI, cs.CY
Key finding The answer to 89% of polls is locatable in the lecture; a poll's response follows its function, not its wording.
arXiv:2609.38959 · Sep 2026 · cs.HC, cs.CY
Key finding 77% of students lean True, so False-keyed items run about thirteen points harder.
arXiv:2609.38969 · Sep 2026 · cs.HC, cs.CY
Key finding A learner's response order is a stable signature (split-half reliability 0.92), unrelated to correctness.
Interested in classroom analytics, collaborations, or the systems behind them? I'd be glad to hear from you.