Akramz•

About

I’m Akram Zaytar, an applied research scientist at Microsoft’s AI for Good Lab. I work on geospatial machine learning: turning satellite imagery into maps that people in agriculture, disaster response, and public policy can use. Before Microsoft, I was at IBM Research.

My recent work covers crop-type mapping and how to evaluate crop maps when field labels are biased, building density and height estimation from satellite imagery, data-efficient land-cover segmentation, and AI agents that label and map imagery. I publish papers on this work and release most of the code as open source.

This site is my personal homepage. The front page is a timeline of what I build and write:

• Code: open-source projects, newest activity first, with weekly contribution counts.
• Papers: journal, workshop, and preprint papers with citation counts from Google Scholar.
• Blog: essays on applied research, AI, rationality, and productivity, plus a series of weekly reading notes.

I also maintain AI for (Public) Goods, a catalog of AI projects built for the public good.

Outside work, I read a lot, run, and try to get a little better every week.

Everything on this site is written by me and reflects my own views, not those of my employer. To reach me, see the contact page.