Rameen Mahmood

Human behavior is difficult to observe continuously at scale, yet it leaves measurable traces in everyday device use. We test whether encrypted smartphone network traffic — a ubiquitous, always-on, passive sensing modality — can passively capture behavioral patterns related to sleep, stress, and loneliness. We model shared behavioral structure using a transformer backbone with per-user adapters, allowing the model to represent both typical individual behavior and deviations from it. To make these representations interpretable, we apply a sparse autoencoder to extract behavioral features corresponding to distinct patterns of activity. We relate these features to sleep disturbance, stress, and loneliness using generalized estimating equations with Mundlak decomposition, separating between-person differences from within-person changes over time. We find that the three outcomes reflect distinct temporal structures: stress is primarily associated with stable between-person differences, loneliness with within-person variation, and sleep disturbance with a combination of both. Notably, these within-person dynamics are not captured by predefined network-traffic features, demonstrating the value of learned representations for longitudinal behavioral sensing. These results establish encrypted network traffic as a viable passive sensing modality, revealing interpretable behavioral dynamics — particularly deviations from an individual's baseline — that are not visible in raw traffic features.

The rapid expansion of IoT devices has outpaced current identification methods, creating significant risks for security, privacy, and network accountability. These challenges are heightened in open-world environments, where traffic metadata is often incomplete, noisy, or intentionally obfuscated. We introduce a semantic inference pipeline that reframes device identification as a language modeling task over heterogeneous network metadata. To construct reliable supervision, we generate high-fidelity vendor labels for the IoT Inspector dataset, the largest real-world IoT traffic corpus, using an ensemble of large language models guided by mutual-information and entropy-based stability scores. We then instruction-tune a quantized LLaMA 3.1 8B model with curriculum learning to support generalization under sparsity and long-tail vendor distributions. Our model achieves 98.25% top-1 accuracy and 90.73% macro accuracy across 2,015 vendors while maintaining resilience to missing fields, protocol drift, and adversarial manipulation. Evaluation on an independent IoT testbed, coupled with explanation quality and adversarial stress tests, demonstrates that instruction-tuned LLMs provide a scalable and interpretable foundation for real-world device identification at scale.

A formative study assessing the feasibility and acceptability of passively capturing encrypted smartphone network traffic from university students to monitor day-to-day digital behavior relevant to mental health. Over a two-week prospective deployment at NYU, 38 students enrolled, 29 provided valid network data, 27 remained active for more than five days, and 25 completed exit interviews. Acceptability was evaluated using the System Usability Scale, NASA Task Load Index, and semi-structured interviews. Beyond feasibility, exploratory analyses link traffic-derived features — timing, intensity, and regularity of use — to aspects of digital behavior relevant to health and daily functioning.

AI-driven applications have become woven into students' academic and creative workflows, influencing how they learn, write, and produce ideas. Conventional survey and interview methods are limited by recall bias and underreporting of habitual behaviors, while ethnographic methods face challenges of scale and reproducibility. We introduce a privacy-conscious approach that repurposes VPN-based network traffic analysis as a scalable ethnographic technique for examining students' real-world engagement with AI tools. By capturing anonymized metadata rather than content, the method enables fine-grained behavioral tracing while safeguarding personal information. A three-week field deployment with university students reveals fragmented, short-duration interactions across multiple tools and devices, with intense bursts of activity coinciding with exam periods — patterns mirroring institutional rhythms of academic life.

RouterSense turns the existing home Wi-Fi router into a passive, low-cost, long-term sensor for behavioral and cognitive health in older adults. By analyzing only metadata from device-router communications — without accessing packet contents — the system captures longitudinal patterns of in-home activity and detects slow drift from a personalized behavioral baseline, with the goal of surfacing early markers of cognitive change.

Version 1.0 of the Assisted Reproductive Technology (ART) Dataset is a multi-modal fertility dataset from treatments performed at the ART Fertility Clinic in Abu Dhabi between 2015 and 2022. The data combine electronic health records and embryo development image sequences captured with the Vitrolife EmbryoScope time-lapse system, providing detailed treatment, morphology, and pregnancy outcome information. The final processed dataset consists of 14,776 embryos from 1,810 patients across 2,500 treatments. The dataset supports the development of machine learning models for automated analysis of embryo development and viability to assist clinical decision-making.