guest@aharoni.dev:~$ █

AMIT AHARONI

ms cs @ columbia // cybersecurity // builder

TRUST NOTHING.

verify everything. ship anyway.

// 00 — whoami

I'm Amit, an MS Computer Science student at Columbia, specializing in cybersecurity. I lead teaching assistant teams in Columbia's CS department, build the apps I wish existed^*, and spend the rest of my time on cryptography, reverse engineering and CTFs.

* I'm usually their first user. sometimes their only one.

// now

Penetration Tester & Security Researcher, Daswani AI pentesting enterprise LLM infrastructureroot-cause analysis of attacks like the XZ Utils backdoor · co-authoring chapters for security publications
Head TA, Columbia CS 30+ TAs across 5 courses · autograder for 400+ students
Board Member, CUCyber Columbia's cybersecurity club
Software Engineer, Reacting Consortium rebuilding the site, front to back
Co-founder & CEO, StudyMesh better classroom collaboration

// before

Head TA, Data Structures in Java COMS 3134 · team of ~20 TAs
Head TA, Advanced Programming COMS 3157 · C, UNIX, systems
SWE Intern, Tetheree Laravel Nova app
Intelligence Officer, Israel Defense Forces lieutenant · special forces platoon medic

// education

MS Computer Science, Columbia Engineering cybersecurity track · 2027
CS undergrad, Columbia University 2026 · Kosoresow Award for TA-ing & service

// off the clock

Half Ironman, St. Pölten training for May 2027 · swim, bike, run
Photography portfolio.amit-aharoni.com ↗
Call of Duty PlayStation, late nights
$1 pizza, NYC unironically the best slice in the city

// 01 — projects

Things I've built.

StudyMesh

Builds study groups from students' schedules, learning preferences and academic strengths using smart matching algorithms.

studymesh.app ↗

Algorithm Trainer

An interactive platform for CS students to master data structures and algorithms through guided exercises, visualizations and step-by-step solutions.

algorithm-trainer.com ↗

EcoCoin

A token and dApp that turn recycled bottles into on-chain rewards, using smart contracts for transparent incentives.

watch demo ↗ source ↗

Miko

Compiles trained neural nets into near-zero-overhead C++ that runs on a Raspberry Pi Pico microcontroller. No TFLite runtime, no heap, just static buffers.

source ↗