Alexander (Alex) Kranias

LinkedIn GitHub X Email

Hi, I'm Alex! I'm a CS student at Georgia Tech.

I'm interested in systems and machine learning.

I love working on hard problems with great people.

Currently I'm

Jan 2025 - Present A research intern in Georgia Tech's SHI Labs. I previously worked on our Generalized Neighborhood Attention project (arxiv.org/abs/2504.16922) - a sparse attention mechanism for vision transformers (MMDiTs/ViTs).

Previously I

October 2025 Made Curserve: a fast and scalable server-side engine for agentic coding (CalHacks 2025). Read Nathan's blog post on it!
May 2025 - Aug 2025 Worked at Apple as a Software Engineer Intern on the MPS Kernels team (Summer 2025). I developed GPU kernels to accelerate on-device inference of MoE LLMs and Diffusion Transformers for Apple's Neural Accelerator.
Aug 2024 - Dec 2024 Worked at AMD as an AI Inference Software Engineer Intern (Fall 2024) on the Triton Kernels team; I developed and optimized new features for our FlashAttention GPU kernels.
October 2024 Created CLaiM: a web app that uses computer vision to automate the filing of property insurance claims after natural disasters strike (CalHacks 2024).
September 2024 Built telepathy: a non-invasive, private silent speech interface between people and machines (HackMIT 2024).
May 2024 - Aug 2024 Developed ROS2 C++ simulation packages of vehicle ad-hoc networks for autonomous mobile industrial robots at Vermeer.
July 2023 - July 2024 Co-founded Sideline: a sports tech startup bringing multimodal vector search and AI scouting to coaching and recruitment. We demoed our product to 17 NCAA colleges, received over 5 million views online, and pitched to NBA leadership.
Jan 2024 - May 2024 Researched novel cache replacement algorithms to mitigate row hammer attacks with Anurag Kar advised by Professor Hyseoon Kim in Georgia Tech's HPArch Lab.
February 2024 Built 3dReal: an iOS app creating NeRFs from networks of mobile phones (TreeHacks 2024).
Aug 2023 - Dec 2023 Assisted with Professor Hyseoon Kim's OMSCS GPU HW/SW course. Wrote a draft of the course's Tiled Matrix Multiplication kernel assignment.
September 2023 Created Bite: a computer vision enabled watch that tracks your daily nutritional intake for you - cal.ai before cal.ai (HackMIT 2023).
Oct 2022 - Sept 2023 Created tools and benchmarks for LLMs at DuckAI: DuckTrack and ARB (accepted to NeurIPS 2023 Workshop).
May 2023 - July 2023 Built a web app that trains and deploys personalized LLMs for financial advisors and investment bankers at Raymond James Financial.
Sept 2019 - Dec 2021 Co-founded Rentathon: a person-to-person rental marketplace for household items.

Writing

December 28, 2025 Compressing a Histogram from 16GB to 2KB.
June 30, 2023 Recreating Hi8.
Always updating... Good (Technical) Reads.

Life