As autonomy becomes a larger and larger part of our everyday lives, a computer’s ability to sense the world around it is becoming increasingly important for developing cutting-edge technologies. My experiences in sensor system design and testing, digital signal / image processing, machine learning, and software engineering provides me with a unique perspective on a computer’s power to understand an analog world.
Sub-Millivolt Electric Field Sensor: Developed data acquisition and signal processing pipeline in MATLAB for highly sensitive electric field sensor. Utilized digital filters, spectral density estimation, and periodogram averaging to detect field in SNR less than 0 dB. Designed and coded MATLAB GUI to provide easy to use, rapid testing environment. Aided in refinement of sensor design.
Circuit Design and Testing: Built, tested, and refined analog filters, three-phase motor drivers, and PID motor controllers for optimizing electric field sensor signal output.
Profiled metabolic factors associated with GABRA5 signaling in medulloblastoma. Implemented dosing and conditioning of human cells lines for drug treatment and molecular assays. Managed all laboratory logistics; including ordering and supply management, solution preparation, and acted as liaison to collaborating scientists. Coordinated with administrative groups to maintain strict laboratory regulations and procedures.
Studied molecular mechanisms of stress granule formation via chromatography and other established biochemical techniques. Investigated the function of Mer1 and its interaction with the spliceosome to determine its role in DNA splicing.
Examined stress-mediated alternative protein translation and its role in cancer development. Researched disrupted circadian rhythm and resulting chromatin remodeling in medulloblastoma cells.
Using Python and OpenCV, developed system that tracks unmarked roads with K-means clustering and morphological transformations on optical images. Implemented techniques present in the literature using image intensity for improved robustness to weather and lighting variations.
Developed Java programming pipeline that encoded 100+ images into a playable video in under 5 minutes using chroma sub-sampling and discrete cosine transform. Allowed for optional video transformations (grayscale, Gaussian blur, color inversion) and output video quality selection.
Investigated and implemented in MATLAB methods for removing gender bias in word embeddings, a commonly used natural language processing technique. Used pre-trained word2vec on Google News text corpera. Tested methods' ability to remove gender bias from gender-neutral words (i.e. doctor, receptionist, etc.) while maintaining gender-specific semantic meaning (i.e. King, Brother, Aunt, etc.). Wrote detailed report and presented findings to peers.
2D convolution is a widely-used, simple, yet computationally inefficient algorithm. Our aim was to explore methods to speed it up by implementing parallelization on GPUs with OpenACC. Coded in C/C++ and tested on Boston University's Shared Computing Cluster. Compared runtimes in serial CPU, parallelized CPU, and parallelized GPU.
GPA: 3.5
GPA: 3.3
Hiker · Self-Described Foodie · Creative Writer · Nerd Culture Aficionado (Level 95 Loremaster) · Aspiring Web Designer · Tutor