Mahmoud Al-Madi
Founder @ Cyberdesk (YC S25)
San Francisco, California, United States · 2 connectionsNot on Unlinked · Imported profile. Is this you? Claim it
Is this about you? Request removal
About
Mahmoud graduated from Rice University with a degree in Electrical Engineering where he built AI & ML solutions at Baylor College of Medicine Neuroscience & Texas Children's Hospital + co-published a paper at the Rice Networks Group. After Rice, Mahmoud built RPA solutions impacting 20K+ employees at a Fortune 100 company, which inspired him to co-found Cyberdesk to build advanced automation capabilities with AI computer use models.
Experience
Mar 2025
Cyberdesk is the computer use agent for developers to automate desktop workflows, backed by YC. We design our agents to follow memorized steps reliably, while falling back to computer use models during unexpected popups. Customers in healthcare, finance, and more are using our agent to automate EHRs, accounting apps, and other desktop apps.
Mar 2024 to Jun 2025
Credera, a global consulting firm, combines transformational consulting capabilities, deep industry knowledge, and Al and technology expertise to deliver valuable customer experiences and accelerated growth across a broad range of industries worldwide.
Jan 2024 to Mar 2025
Experience Streamline (experiencestreamline.com) – swim team search and lesson booking platform. - I built this with the intention of licensing it to a founder working in the club sports space. While the site currently contains only dummy data since it hasn't been commercialized yet, I did license it and it’s fully functional and production-ready. In terms of implementation, I prioritized speed and practical execution: - Frontend: React with libraries for complex components like a calendar-based dashboard - Backend: Personally configured AWS EC2 server - Frontend deployment & Auth: Firebase for data/auth, AWS Amplify for serverless frontend deployment TextCopilot (textcopilot.com) – (iOS & Android) chatbot app built as a passion project. - I designed and built the frontend from scratch in React Native, including custom icons made in Figma. I implemented a custom memory system, featuring a Retrieval Augmented Generation (RAG) algorithm. - On the backend, I managed my own PostgreSQL database on Amazon RDS, configured EC2 instances in autoscaling groups behind a load balancer, and built a full CI/CD pipeline connected to my GitHub repo. I personally handled all EC2 infrastructure and backend service setup. - After launch, I onboarded 100 users from friends and family. I chose not to pursue it commercially -- it was built purely to explore and learn. Recically – multiplatform (iOS & Android) social meal planning app - built out of a passion for collaborative meal planning. - Front-end: React-native - Backend: AWS EC2, Amazon RDS - Auth: Firebase
Jan 2022 to Feb 2024
Developed a random‑forest model that forecasts mouse mortality within five minutes of lab readings at up to 84% recall, using real‑time respiratory waveforms and ECG data. Engineered hundreds of candidate features, pruned them to 100 variables explaining >90% of variance using an iterative feature elimination algorithm, and leveraged bayesian hyperparamter tuning to optimize performance. Other fun stuff I did: - Extracted QRS wave features from mouse ECG data by designing algorithms using PyWavelets in Python - Identified statistically significant survivability factors using Kaplan-Meier Survival Analysis and Cox PH regression - Fourier analysis of mice respiratory and cardiac data Implemented Principle Component Analysis (PCA) based plotting and clustering to find hidden patterns and clusters - Data mining pipeline includes overrepresentation analysis to identify overrepresented genotypes and features in clusters to help classify clusters - Classified clusters to be used to build and inform a diagnostic machine learning model to detect anomalous breathing
May 2023 to Nov 2023
Co‑published a paper on disrupting mm‑wave acoustic eavesdropping via programmable silicon metasurfaces. Built a MATLAB pipeline that converts mm‑wave radar returns into earpiece audio by tracking micrometer‑scale phone‑case vibrations from up to 2 m away. Programmed dynamic voltage patterns on the metasurface so that, when mounted behind the phone, it injects false signals (such as the Declaration of Independence in one of the experiments) and misleads the radar’s eavesdropping. Accepted into IEEE Symposium on Security and Privacy 2025 (IEEE S&P 2025) and Optica [Vol. 11, Issue 8, pp. 1113-1114 (2024)]
Aug 2022 to May 2023
Developing Named Entity Recognition algorithms with NLP transformer models along with keyword and dependency-chain matching to automate data entry processes--hence boosting nurse productivity
Sep 2021 to Dec 2022
Hosted weekly office hours Graded weekly problem sets
May 2022 to Aug 2022
Hosted weekly research talks to give program participants opportunities to present and discuss their research Organized, hosted, and facilitated social events throughout the duration of the program
Jan 2021 to Aug 2021
Integrated DPDK libraries in Agora, real-time MIMO baseband processing software, to reduce computational latency Debugging C++ programs in Agora to ensure their compatibility with the newly integrated DPDK libraries Compiling programs using CMake to build programs in Agora more efficiently in a Linux environment
Education
Jun 2025 to Sep 2025
Bachelor of Science - BS, Electrical and Computer Engineering