Qalab Hassnain AghaCo-Founder & CTO
ResumeBook a call

Research

Research and publications

I study what makes AI hold up outside the lab: on edge devices, in medical imaging and in voice agents. 3 published papers and 2 under review or submitted, with open data and code wherever I can share them.

Google Scholar ORCID ResearchGate

  • PublishedInternational Journal of Advanced Research, 3(7) · November 2026

    Edge AI-Enabled Intelligent Healthcare Framework Using Lightweight Deep Learning, Multimodal Sensor Fusion, and Real-Time On-Device Analytics for Continuous Patient Monitoring and Early Health Risk Detection

    First and corresponding author

    A framework for continuous patient monitoring that runs on edge hardware: wearable and bedside signals fused by attention, a compact network shrunk by pruning, distillation and 8-bit quantization, and federated learning so raw physiological data never leaves the device.

  • Under reviewComputer Speech & Language

    “Wait, Cancel That”: Release Monotonicity and Cancellation Semantics for Streaming Voice-Agent Actions

    Sole author

    Voice agents often act before the user finishes speaking, so a spoken correction can arrive after an irreversible message has already gone out. The paper proves that holding irreversible effects until the agent’s own end-of-turn decision never adds a superseded effect and adds no acknowledgement latency, then checks it across 12,348 simulated runs with zero violations. Code and every run are open source.

  • PublishedJournal of Global Social Transformation, 2(10) · October 2026

    Feature Selection and Model Optimization for Accurate Classification of Complex Data

    First author

    How much do feature selection and tuning actually buy you? On the UCI Adult Income dataset (48,842 records), mutual-information feature selection halved the predictors from 14 to 7, while Gradient Boosting accuracy moved from 86.9% to 85.7%: half the inputs for a 1.2-point accuracy cost.

  • PublishedInternational Journal of Advanced Research, 3(6) · October 2026

    Beyond Automation: Generative AI’s Influence on Human Decision-Making, Trust, and Accountability in Intelligent Cybersecurity Systems

    Co-author, 4 of 7

    A framework for how generative AI shapes human decisions, trust calibration and accountability in cybersecurity, with seven testable propositions and a risk-sensitive governance model.

  • SubmittedComputer Methods and Programs in Biomedicine

    What Transfers From Synthetic to Real Scalp Dermoscopy? A Pre-Deployment Component-Level Evaluation for Handheld Trichoscopy

    Sole author

    A component-by-component evaluation of image analysis for a handheld scalp camera, built before device images existed. Components trained on synthetic images failed on real dermoscopy. Which segmentation method looked best reversed depending on whether the test labels had been seeded by one of the methods, evidence of annotation bias. Optical scale estimated from hair width was off by a median of 37% against printed scale bars. Every result is reproducible from openly released data and code.

Have an AI system that needs to survive production?

Tell me what you’re building. In 30 minutes you’ll get an honest read on the architecture, the risks and what I would build first. No commitment, no sales pitch.

Book a 30-min call Send a messagesupport@qalabagha.com