About

My work in machine learning has been shaped by problems where signals are noisy, observations are incomplete, and reliable structure has to be recovered under real-world constraints.

That perspective developed through work on physiological signals, interactive textile sensing, and naturalistic behavioral data, but the underlying questions have remained strikingly consistent: what can be learned from sparse or difficult observations, how stable that learned structure remains across users and conditions, and how far representation design and temporal modeling can go in compensating for what the data or hardware cannot provide.

I am especially drawn to problems where modeling, sensing constraints, and deployment realities all matter at once. Those challenges first became clear to me in constrained interactive systems, but they have continued to shape how I think about machine learning much more broadly.

How I Think About ML Problems

I approach machine learning as a systems problem, not a modeling problem in isolation.

That means the questions I ask are not only which model performs best, but whether the evaluation actually tests generalization, whether the representation holds up when users or environments change, and what deployment requires of the full preprocessing and modeling pipeline.

To support this kind of work, I developed BioHCI, a modular time-series learning framework that evolved from early physiological modeling through capacitive touch sensing into reusable infrastructure for structured human-generated signals. BioHCI reflects an important part of how I work: I do not think of modeling, preprocessing, evaluation, and deployment as separate stages loosely stitched together, but as interdependent parts of the same learning system.

That perspective tends to push toward choices that hold up outside the lab. It also makes the work more interpretable, not just to reviewers, but to the engineers, clinicians, and designers who need to act on what the model produces. I am drawn to modeling decisions that are analytically defensible, legible to domain experts, and implementable on the systems that actually exist.

I often treat predictive tasks not only as endpoints, but also as probes of representation quality: not just whether a model performs well, but whether it captures structure that is stable, meaningful, and likely to transfer across users, environments, and conditions.

Research Background

I earned my PhD in Computer Science at Drexel University, where my research spanned both physiological signal modeling and interactive textile sensing. My early work included fNIRS data, which helped build my foundation in signal processing, feature construction, and machine learning for noisy, temporally structured human data. That work also laid the foundation for BioHCI, the modular framework I later expanded across multiple sensing domains. It also introduced questions that would stay with me: how temporal resolution affects what can be learned, how representation choices shape performance, and how much useful structure can be recovered from difficult biological signals.

My dissertation focused on the real-world interactivity potential of minimalistic knitted sensors at the intersection of machine learning and human-computer interaction. Building on my earlier experience with physiological signals, I developed computational approaches for sparse textile-based sensing systems that could recover human intent through touch and gesture without sacrificing manufacturability, robustness, or usability.

Following my PhD, I completed a postdoctoral fellowship at the Center for Autism Research at CHOP, where I developed computational frameworks for behavioral and physiological time-series data in naturalistic settings. Across these stages of my work, a consistent thread has been recovering stable, meaningful structure from noisy, human-generated signals and evaluating whether that structure remains informative across users, environments, and conditions.

What I Bring

The specific domains of my published work — physiological signals, textile sensing, and behavioral analysis — are narrower than the range of machine learning problems I can contribute to. What transfers is the reasoning, along with technical strengths in representation learning, temporal modeling, cross-subject evaluation, and end-to-end machine learning for structured and sequential data:

  • designing representations for signals that are sparse, noisy, temporally structured, and variable across users and conditions
  • building temporal models and evaluation frameworks that test for generalization rather than only optimizing held-out accuracy
  • making architecture decisions with deployment and implementation constraints in mind from the start
  • working across research, engineering, and applied contexts without losing rigor in either direction
  • Technically, this work has involved Python, PyTorch, scikit-learn, NumPy/SciPy, sequence models including LSTMs and CNNs, and edge deployment on NVIDIA Jetson platforms.

I am comfortable in research settings where the right model, representation, or evaluation strategy is still an open question, and in engineering settings where the challenge is making a trained system work reliably in practice. The work I find most interesting often sits at the boundary between those two.

Collaboration


Across my work, meaningful collaboration has taken different forms: translating user study findings into concrete modeling requirements with HCI researchers, making evaluation design legible to clinicians who needed to trust the output without understanding the pipeline, and working with engineers on the gap between a system that runs on embedded hardware and enough signal to understand human intent.

What those experiences share is that the technical and the contextual have to develop together: domain knowledge shapes which modeling choices are defensible, and modeling constraints shape which research questions are actually answerable.

Explore further

Research: Questions, methods, and findings across three research domains, with the modeling reasoning made explicit.

Systems: BioHCI and the end-to-end pipeline architecture behind the work, from signal acquisition through embedded deployment and robustness evaluation.

Publications: Full publication list organized by research area, with abstracts, PDFs, and slides.

CV | GitHub: BioHCI framework (10,000+ lines, Python) | Google Scholar | LinkedIn