• Developed a Python library leveraging generative AI to design synthetic DNA—currently used by two genetic engineering labs.

• Optimized model to train 16x faster through distributed data parallelism and mixed precision; tailored model for wide-format DNA sequence data and small-scale feature extraction.

• Demonstrated AI generated synthetic regulatory DNA’s impact on gene expression in mouse embryonic stem cells, advancing applications in gene regulation research.

• Developed a deep learning model enabling prosthetic limbs to recognize and execute up to 65 distinct gestures, a significant advancement over the standard two-gesture (open/close) systems on the market.

• Co-authored and presented at IEEE ICRA 2024: https://ieeexplore.ieee.org/document/10610638

• Achieved the highest recorded classification accuracy (98.03%) on a benchmark EMG dataset, leveraging a novel multimodal transformer deep learning model

• Designed and developed a server (async + multiprocessing) to connect testers (Clients, GUIs, Automation scripts) with a hardware simulation of a surgically implanted urology device. Can handle 100+ clients, coordinating multiple physical machines.

• devices simulating pressures / temperatures with <10 ms latency.

• Rewrote 20K+ lines of legacy code into a 2K line Python maintainable service layer (100% test coverage); stress tested with concurrent 4,000 requests from 40 clients to ensure safe, repeatable device validation.

• Developed a haptic vest enabling blind / visually impaired individuals to "see" their environment through tactile feedback, conveying the location and distance of obstacles without relying on a cane. Improved traversal speeds by 1.4x and reduced the number of collisions by 2x.

• Designed and built custom electronics using Altium and soldering; the final design ensured high contact, comfort, and washability, with concealed wiring for a fashionable and practical design.

• Simulated real-world scenarios using a VR environment to test the vest's performance in crowded, chaotic spaces like subway stations, addressing the challenge of limited blind tester availability.