Oct 2023 – Mar 2026 · 2 yrs 5 mos · Jersey City, New Jersey, United States

• I am a member of the AI, CV, and Robotics Software team at Oishii.
• I teach our robot arms to detect and harvest the very best, ripest strawberries.
• Highlights:
• Strawberry detection and ripeness estimation
• Strawberry localization in 3-D space and stem point estimation
• Metrics generation to analyze harvesting outcomes, performance, and future yield prediction

Apr 2022 – Oct 2023 · 1 yr 6 mos

• Developed high-accuracy TensorFlow CNN model for tote bale arm state classification, leveraging Azure Machine Learning. Streamlined ground truth image collection and preprocessing through tooling and pipelines for rapid model training iterations.
• Implemented Dockerized Python apps to deploy deep learning solution, generate telemetry data, and monitor disk utilization on a central server. Designed dashboards in Azure Data Explorer to provide real-time vision system performance insights. Implemented anomaly detection algorithms to identify out-of-distribution samples in production data.
• Collaborated with site engineers, resolving vision system issues within 4 hours and minimizing downtime.
• Contributed significantly to the vision library for the static workstation product, being one of three core team members. Reponsible for Jenkins CI infrastructure code, Artifactory integration, and actively addressed bug fixes and feature additions.
• Optimized the vision system provisioning process by simplifying instructions, incorporating user feedback mechanisms, and refining procedures for calibration, white balance, exposure, focus, and polarizer adjustment. This resulted in a reduction in setup time and improved scalability for future deployments.

Jun 2020 – Apr 2022 · 1 yr 10 mos

• Developed a proof-of-concept robot detection and distance estimation algorithm using robot on-board cameras in C++ and performed a study to determine the feasibility of running the vision process in real-time.
• Designed and implemented Qt-based hierarchical state machines for the Hansel Portal auto-dispense robotic system. Performed site support and bug fixes as the product was launched in a production environment.
• Supported the evolution of the Portal product as needs arose to integrate it with new UI and backend systems. Was responsible for designing, implementing, validating, and documenting features through unit testing and UML diagrams.

Aug 2019 – Dec 2019 · 4 mos · North Billerica, Massachusetts, United States

• Developed a working prototype with Python and OpenCV for object detection inside the Alphabot automated storage and retrieval system (ASRS) using robot on-board cameras.
• Built an extensible video processing framework for rapid prototyping - useful to plug-and-play new algorithms and observe the results.
• Worked in a cross-functional team employing Agile development techniques.

May 2019 – Aug 2019 · 3 mos · Greater Boston Area

• In this role with the Hardware Description Language (HDL) Coder team, I added support for the half-precision floating point data type in MATLAB.
• In addition to supporting more features, I also researched and created examples for the MATLAB documentation.
• As an EDG intern, I had the opportunity to answer technical support questions from MathWorks customers using various products, such as the Image Processing Toolbox, Symbolic Math Toolbox, and Deep Learning Toolbox.

Jan 2019 – May 2019 · 4 mos · Greater Atlanta Area

• As a peer tutor, my role was to assist Georgia Tech student athletes with mathematics courses.
• I lead tutorial sessions, either one-on-one or in small groups, to illustrate and explain mathematical concepts and work through practice problems for students with diverse academic backgrounds.

Aug 2018 – Dec 2018 · 4 mos

• In this directed research project under Dr. Ghassan Al-Regib, I worked on identifying fluid-filled regions in Optical Coherence Tomography images using deep learning techniques.
• I implemented a deconvolution CNN architecture to isolate regions of edema fluid in OCT scans using PyTorch.
• In MATLAB, I utilized graph search, texture similarity, and thresholding techniques to prepare ground truth pixel-level labels in an unsupervised manner.

May 2017 – Dec 2017 · 7 mos · Singapore

• I worked as a research intern in the Ophthalmic Engineering and Innovation Laboratory in the Biomedical Engineering Department of the National University of Singapore. My work involved developing next generation image processing techniques to compensate (remove attenuation) in optical coherence tomography images of the optic nerve head. This work serves as a precursor to automated detection and diagnosis of glaucoma, a pernicious eye disease.

Nov 2016 – Jan 2017 · 2 mos · Bangalore

• I developed a computationally efficient linear filtering approach in MATLAB to invert banded Toeplitz matrices in Θ(n) complexity, a decrease from the standard Θ(n^2). Toeplitz matrices are often encountered in digital signal processing problems and, as such, have tremendous engineering significance.

May 2016 – Jul 2016 · 2 mos · Bangalore

• The objective of my summer research internship under Dr. Chandra Sekhar Seelamantula was to create an Android application for segmentation and outlining of the optic disc and optic cup in retinal fundus images. The cup-to-disc ratio being a powerful indicator of glaucoma, this project has immense medical significance. Developing an Android application for this purpose eliminates the requirement of buying expensive medical equipment, and is thus a step in making glaucoma detection as cost-effective as possible.