Developing investigations into learning, behavioral evidence, engineering systems, and decisions under uncertainty.
My research begins with problems where the outcome cannot be observed directly, the available signals are incomplete, and decisions still need to be made.
These entries represent the current state of my reasoning rather than final unquestionable conclusions. Each investigation can change as new evidence, methods, or contradictions appear.
A central direction across this work is understanding what behavioral signals can tell us when the phenomenon we care about—such as learning, credibility, or understanding—is latent and cannot be observed directly.
What if software acknowledged you? Most interfaces silently execute commands. TalkUI tests a different interaction model: the interface briefly communicates what is happening before, during, or after an action. https://talkui.research-prototype.nirajchaurasiya.com/
How should interfaces communicate consequential system information so that users notice, understand, and act without creating unnecessary distraction?
The traces left by users, whether conscious or unconscious, may differ across domains and must be measured probabilistically with care.
A completed academic investigation using EPA air-quality data to examine the difference between observing environmental conditions through monitoring and extending those observations through modeling.
Is there a significant difference in reported air-quality values between observations derived from Monitor Only data and those derived from Monitor & Modeled data?
An investigation into how learning can be inferred when the internal process itself is not directly observable and feelings of clarity, attention, or engagement may provide weak evidence.
How can we determine whether learning has occurred when learning itself cannot be observed directly?