Brought up hardware accelerator for next-gen radar chip on self-driving delivery robot. Implemented accelerated routines for digital signal processing algorithms and optimized on-chip radar DSP pipeline (mostly pure C with a tiny bit of assembly).

Developed a rigid-body grasping simulator in PyBullet and used synthetic demonstration data to train a planner for robotic assembly tasks.

Onboard and embedded communications infrastructure for radar driver module on self-driving delivery robot. Improved sensor data throughput and firmware upgrade speeds (mainly C, C++, Python, CAN-FD, Ethernet).

Core member on the WATonomous student design team working to build a level 4 autonomous vehicle for the SAE AutoDrive Challenge.• Working on optimizing the existing radar driver implementation in C++ and ROS by applying more intelligent filtering methods, mitigating interference, and improving the existing point cloud visualization

Designed real-time, multithreaded video analysis application in C++ for low power embedded platform (i.MX8M Mini board with 4xA53 ARM cores)• Refined low-latency multi-object tracking algorithms in OpenCV, improving tracking accuracy by 230%• Improved detection accuracy on pedestrian rich images by performing transfer learning on MobileNet SSD network with Tensorflow, and quantized model with TfLite to optimize for embedded use• Developed internal library to accelerate CPU based inference using ArmNN and integrated it into existing toolchain with Bitbake, reducing inference time by 460ms• Automated CI pipeline using Jenkins to validate tracking performance against global MOT benchmark

Worked across embedded software engineering teams to develop, test, and debug firmware for use on ruggedized single board computers • Implemented, validated, and released features for U-Boot bootloaders to perform PBIT tests, control onboard sensors over UART using IPMI, and update firmware images over JTAG interface (C, KBuild) • Designed tool utilizing RSA based encryption in Python to securely transfer payloads to customers • Received award for driving debug and validation efforts of VxWorks & Yocto Linux BSPs, BIOS, and bootloaders

Worked on Design Verification & Validation team to architect an internal tool for simulating DMA performance of NXP's next-gen S32V SoC for ADAS. • Analyzed simulation results and register level waveforms to identify bugs in SystemVerilog RTL resulting in 2.73x improvement in DMA bandwidth.  • Automated data visualization from overnight regression tests to proactively diagnose performance issues.

Worked on project to migrate the Home Depot Canada site to React• Developed React components for an internal front-end library• Automated bilingual content translation for global navigation flyout by building a web scraping tool that found the equivalent product description in French given an English product group.• Developed Python tool to conduct Market Basket Analysis on e-commerce consumer data• Used Gulp to automate icon generation and optimize distribution build processes

The Oxford Learning method believes that there's more to tutoring than helping kids master content - what's more important is teaching them how they can master that content by themselves in the future. As an academic tutor, I worked one-on-one with students from KG - Grade 12, imparting study skills and expertise in Reading, Writing, Math, Physics, and Chemistry. My ability to explain challenging and technical Math & Science concepts, combined with an infectious drive to pass on to students my enthusiasm for learning, made me a highly sought after tutor within the centre.

Worked closely with school principal in an administrative capability while also supporting teachers in classroom activities.