Working on ComputeCore, CoreAVI’s SafeAI Library – A neural network inference engine built on Vulkan SC• Implemented and debugged multiple neural network operators such as split, concatenate, and not using C code, Vulkan, and GLSL shaders to execute computations on the GPU• Developed APIs to enable applications to acquire ownership of ComputeCore’s buffers and synchronize memory accesses using semaphores, eliminating the need for memory transfer between GPU and CPU• Automated neural network tests using Python and PyTorch scripts to improve QA testing efficiency• Updated ComputeCore’s software test plan by collaborating with the Research, QA, and Engineering teams

• Responded to 1000+ technical customer support inquiries to resolve a variety of e-bike issues, approved hundreds of warranty parts (eg. throttle, controller), and provided technical help to many custom EV projects• Wrote 18+ instructional articles about DIY e-bike conversion kits (goldenmotor.bike/support/instructions)• Onboarded and trained a new technical support specialist to take over my role

• Managed 120+ Windows and Linux devices and servers using Microsoft Azure and Intune• Wrote 15+ articles documenting IT processes and procedures on GitHub such as how to re-image Linux and Windows machines, build new PCs, enable BitLocker encryption, onboard/offboard users, etc.• Resolved a variety of user IT issues related to connectivity, permissions, malfunctioning equipment, etc.

• Developed a Telegram bot using Python to deliver stock market alerts to a subscription-based trading channel• Deployed the bot using AWS Lambda and an on-prem server, ensuring low latency and 24/7 uptime• Collaborated closely with a non-technical client throughout the software development life cycle, iterating over bot functionality and performance to align with business requirements