<aside> <img src="/icons/suspension-bridge_lightgray.svg" alt="/icons/suspension-bridge_lightgray.svg" width="40px" /> About I currently live in SF and work at Affirm as a software engineer. I’m currently focused on building better tools for ML (feature stores and ETL pipelines) and understanding how infrastructure (databases, data warehouses, containers) shapes software design.

Outside of software, I’m interested in history (computing, technology), math, and games.


<aside> <img src="/icons/link_lightgray.svg" alt="/icons/link_lightgray.svg" width="40px" /> Links Github: Rostifar Email: [email protected]


<aside> <img src="/icons/branch-create_lightgray.svg" alt="/icons/branch-create_lightgray.svg" width="40px" /> Projects

Fractalscape (2017)


Fractalscape is a VR application for visualizing 3D fractals built by Lance Bowley and myself. The app has received 7000+ downloads and was featured on the frontpage of the Oculus app store. Here is a short promo.

Taglets (2022)

Taglets is a system for building data-efficient classifiers using auxiliary data sources; as an example, ConceptNet can be used to improve image classification performance by associating imagery that share high-level concepts with the same label. This work was published in MLSYS 2022. The technique is similar to ensembling or mixture of experts.

Quantile Estimation in Streams (2021)

I wrote a survey on sublinear algorithms for approximating quantiles in a data stream as part of my CS 1951w final project. You can find the paper here.