Skip to content
View njaric03's full-sized avatar

Block or report njaric03

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
njaric03/README.md

The bus network a graph policy proposes for Novi Sad, drawn over the city street grid

MSc student in computer science at the University of Novi Sad. In 2025 I was an applied science intern at Microsoft, on a team getting language models to run on constrained hardware. My own part was personalization, inferring a user's writing style and profile from on-device data, which ended up as a patent application I co-authored.

Most of what is pinned here is applied machine learning on spatial data. Population estimated from satellite imagery and building footprints, a building typology for Novi Sad learned from clustering instead of defined up front, a graph policy for transit network design that belongs inside local search rather than in place of it.

Outside of that I like measuring public data about Serbia, usually where practice and the regulation part ways: burn scars, air quality, retail prices, stuff like that.

Pinned Loading

  1. tndp-novi-sad tndp-novi-sad Public

    Transit Network Design with a GATv2 policy trained by REINFORCE: benchmarked on Mandl and Mumford instances, applied to Novi Sad.

    Python

  2. novi-sad-building-typology novi-sad-building-typology Public

    Building typology of Novi Sad: 100,000+ footprints clustered on geometry and surroundings, from Overture Maps, Sentinel-2 and OpenStreetMap.

    Jupyter Notebook

  3. population-from-satellites population-from-satellites Public

    Population estimation for 4,721 Serbian settlements from Sentinel-2 imagery and building footprints, validated against the 2022 census.

    Jupyter Notebook