# Akram Zaytar

> Applied research scientist at Microsoft's AI for Good Lab working on geospatial machine learning: satellite imagery to maps for agriculture, disasters, and public goods. This is his personal site: open-source code, papers, and blog posts.

Pages: [About](https://akramz.space/about/) · [Contact](https://akramz.space/contact/) · [Privacy](https://akramz.space/privacy/) · [Sitemap](https://akramz.space/sitemap.xml) · [RSS](https://akramz.space/feed.xml)

## Papers

- [Crop Mapping Evaluation under Sampling Bias](https://akramz.space/assets/papers/2026-crop-map-evaluation-sampling-bias.pdf): arXiv, 2026
- [Survey Protocol Cards for Crop Maps](https://akramz.space/assets/papers/2026-survey-protocol-cards-crop-maps.pdf): IEEE GRSL, 2026
- [GeoAI Agency Primitives](https://akramz.space/assets/papers/2026-llm-tools-geospatial.pdf): ICLR workshop, 2026
- [Core-Set Selection for Data-Efficient Land Cover Segmentation](https://akramz.space/assets/papers/2026-smaller-datasets-same-performance.pdf): IEEE Access, 2026, 3 citations
- [TEMPO: Global Temporal Building Density and Height Estimation from Satellite Imagery](https://akramz.space/assets/papers/2025-detecting-buildings-3-months.pdf): arXiv, 2025, 4 citations
- [Bootstrapping Rare Object Detection in High-Resolution Satellite Imagery](https://akramz.space/assets/papers/2024-labeling-rare-objects-satellite.pdf): ICLR workshop, 2024
- [Satellite Imagery Noising with Generative Adversarial Networks](https://akramz.space/assets/papers/2020-satellite-imagery-noising-gans.pdf): IJCINI, 2020, 6 citations
- [Sequence to Sequence Weather Forecasting with Long Short-Term Memory Recurrent Neural Networks](https://akramz.space/assets/papers/2016-weather-forecasting-lstms.pdf): IJCA, 2016, 420 citations

## Code

- [AI for (Public) Goods](https://akramz.space/ai4g/): Catalog of AI for public-good projects.
- [dotfiles](https://github.com/Akramz/dotfiles): My OS + coding-agent setup. ★ 2
- [wispr-cli](https://github.com/Akramz/wispr-cli): Wispr Flow meetings & notes for agents.
- [PaPPI](https://github.com/microsoft/PaPPI): Audit crop-map accuracy when field labels are biased. ★ 6
- [html-explain](https://github.com/Akramz/html-explain): Explain systems as diagrams you annotate. ★ 2
- [nxmndr](https://github.com/microsoft/nxmndr): Geospatial AI inference server. ★ 4
- [pytorch-cloud-geotiff-optimization](https://github.com/microsoft/pytorch-cloud-geotiff-optimization): Fast GPU training on cloud GeoTIFFs. ★ 60
- [rice-irrigation-mapping-s1s2](https://github.com/microsoft/rice-irrigation-mapping-s1s2): Rice irrigation mapping from Sentinel-1/2. ★ 13
- [template](https://github.com/Akramz/template): Agent-ready Python research repos. ★ 1
- [mpc](https://github.com/Akramz/mpc): Let AI agents read Planetary Computer data. ★ 1
- [ccai-food-security-tutorial](https://github.com/Akramz/ccai-food-security-tutorial): Climate Change AI food-security tutorial. ★ 4
- [buildings](https://github.com/microsoft/buildings): Global building density & height maps. ★ 85
- [Sims](https://github.com/microsoft/Sims): Interactive geospatial feature discovery. ★ 51
- [AFRIC_GeoAI_workshop](https://github.com/Akramz/AFRIC_GeoAI_workshop): AFRIC 2 GeoAI workshop notebooks.
- [data-centric-satellite-segmentation](https://github.com/microsoft/data-centric-satellite-segmentation): Sample EO training data under a budget. ★ 57
- [building-damage-assessment](https://github.com/microsoft/building-damage-assessment): Map post-disaster damaged buildings. ★ 165
- [vllm-satim-labeling](https://github.com/GeoAIAfrica/vllm-satim-labeling): Vision-LLM satellite labeling challenge. ★ 1
- [interactive_geospatial_mapping](https://github.com/GeoAIAfrica/interactive_geospatial_mapping): AMLD 2024 geospatial workshop. ★ 21
- [indaba-pracs-2023](https://github.com/deep-learning-indaba/indaba-pracs-2023): Deep Learning Indaba 2023 practicals. ★ 141
- [satellite-data-manipulation-for-ml](https://github.com/Akramz/satellite-data-manipulation-for-ml): Tiles + polygons to a CV dataset. ★ 4
- [jupyter_plz](https://github.com/Akramz/jupyter_plz): Copilot for Jupyter notebooks. ★ 29
- [end-to-end-gee](https://github.com/Akramz/end-to-end-gee): End-to-end Google Earth Engine, in notebooks. ★ 78

## Blog

- [How to Write?](https://akramz.space/how-to/2026/07/11/how-to-write.html): 2026-07-11
- [Weekly reads #15](https://akramz.space/notable-reads/2025/07/13/weekly-reads-create-for-yourself.html): 2025-07-13
- [Weekly reads #14](https://akramz.space/notable-reads/2025/06/29/weekly-reads-life-principles-belief.html): 2025-06-29
- [Weekly reads #13](https://akramz.space/notable-reads/2025/06/22/weekly-reads-research-free-markets.html): 2025-06-22
- [Weekly reads #12](https://akramz.space/notable-reads/2025/06/14/weekly-reads-multidisciplinary-advantage.html): 2025-06-14
- [Weekly reads #11](https://akramz.space/notable-reads/2025/06/07/weekly-reads-early-consciousness.html): 2025-06-07
- [Weekly reads #10](https://akramz.space/notable-reads/2025/05/24/weekly-reads-air-light-and-human.html): 2025-05-24
- [Weekly reads #9](https://akramz.space/notable-reads/2025/05/09/mindfulness-leverage-and-forward.html): 2025-05-09
- [Weekly reads #8](https://akramz.space/notable-reads/2025/04/12/freedom-focus-and-the-folly-of-control.html): 2025-04-12
- [Weekly reads #7](https://akramz.space/notable-reads/2025/03/22/project-management-Moloch-Beyond-Programming.html): 2025-03-22
- [The Hands-on Guide to Applied ML Research](https://akramz.space/how-to/2025/03/15/on-applied-research.html): 2025-03-15
- [Weekly reads #6](https://akramz.space/notable-reads/2025/03/15/llm-coding-ai-predictions-scorpion.html): 2025-03-15
- [Weekly reads #5](https://akramz.space/notable-reads/2025/03/10/ai-purpose-research-boring-tech.html): 2025-03-10
- [Weekly reads #4](https://akramz.space/notable-reads/2025/03/01/notable-reads.html): 2025-03-01
- [Weekly reads #3](https://akramz.space/notable-reads/2025/02/22/notable-reads.html): 2025-02-22
- [Weekly reads #2](https://akramz.space/reads/2025/02/15/notable-reads.html): 2025-02-15
- [Weekly reads #1](https://akramz.space/reads/2025/02/08/notable-reads.html): 2025-02-08
- [Weekly reads #0](https://akramz.space/links/reading/2025/01/25/notable-reads.html): 2025-01-25