experience
My research, the things I've built, and where I've worked.
Computational optical imaging for non-invasive, equitable diagnostics.
- Smartphone nailfold capillaroscopy: designed a green-LED back-illuminated optical relay (25 lp/mm, resolving 20 μm capillary loops) for non-invasive hemoglobin estimation.
- Skin-tone-aware acquisition: partnered with Samsung Mobile engineers on an exposure- and focus-controlled pipeline that keeps capillary contrast usable across skin tones, now the lab default.
- Light-transport modeling: building Monte Carlo forward models that map image contrast to capillary depth and hemoglobin concentration.
Efficient and reliable medical vision models.
- Adaptive inference: led a framework for medical Vision Transformers that decides, per image, when to reduce tokens and when to exit early, preserving diagnostic accuracy at a fraction of the compute. MIDL 2026
Co-founder · OcuSound 2023 – Present
A low-cost at-home tonometer for glaucoma monitoring.
- Device & signal processing: drove the bill of materials below $100 per unit and wrote the pipeline that extracts pressure-correlated features from acoustic and infrared signals.
- Clinical study & traction: led an IRB-approved feasibility study at a partner clinic in India, raised $25K in non-dilutive funding, and won 1st place at the Hopkins New Venture Challenge. ARVO 2026
R&D New Product Development Intern · STERIS Endoscopy 2024
- Redesigned SolidWorks test fixtures, cutting prep time by 40%, and built an automated test rig that reduced manual data-tracking by 80% during reliability testing.
Engineering Intern · Clear Guide Medical 2023 – 2024
- Automated CNN-based needle segmentation in ultrasound to improve needle-track overlay, and built a LaTeX and Python documentation toolchain that cut audit-prep time by 30%.
Selected projects
- Deep learning for melanoma diagnosis on darker skin (2025): style-transfer augmentation to counter skin-tone imbalance in a dermoscopy dataset, improving accuracy on the underrepresented subset.
- Sleep-apnea prevention device (2022 to 2023): capstone device delivering a controlled jaw-thrust within 60 seconds, iterated across three prototypes and 25 user interviews.
Leadership & service
- Treasurer, The MedTech Network, JHU (2023 to present)
- Promotions Co-Lead, MedHacks, JHU (2022)
Skills
Python (PyTorch, NumPy, OpenCV, timm, scikit-learn), C/C++, MATLAB, LaTeX · Vision Transformers, CNNs, token reduction, early-exit inference, dataset-bias analysis · Monte Carlo light-transport modeling (mcxyz, MCML) · SolidWorks, circuit design, 3D printing, Raspberry Pi and Arduino · smartphone camera characterization and optical, acoustic, and infrared sensing.