Ahmad Al Badawi

Making encrypted AI practical.

I build the cryptography that lets AI run on data it never gets to see. Co-author of PALISADE/OpenFHE, former technical lead on the DARPA DPRIVE 12 nm FHE accelerator, and principal investigator on private LLM inference under fully homomorphic encryption.

Senior Cryptography Scientist at Duality Technologies.

Portrait of Ahmad Al Badawi

Simulators

  1. A Step-by-Step Simulator for the NTT in FHE

    An interactive simulator that replays the real FHE NTT one butterfly at a time: recorded traces and the six places where the production code differs from the textbook.

Writing

  1. The Math of Slot Rotation in BGV and BFV

    How slot rotation works in BGV and BFV homomorphic encryption: automorphisms on cyclotomic rings, Galois groups, and worked examples on packed SIMD ciphertexts.

  2. SIMD Packing in BGV/BFV FHE Schemes

    A deep dive into SIMD packing in BGV and BFV homomorphic encryption: the vector/polynomial isomorphism, CRT over cyclotomic rings, and a Python walkthrough.

Selected papers

  1. Cryptology ePrint2026

    SoK: Private LLM Inference using Approximate Homomorphic Encryption

    Al Badawi, A., Alexandru, A., Polyakov, Y., & Vaikuntanathan, V.

  2. WAHC2022

    OpenFHE: Open-Source Fully Homomorphic Encryption Library

    Al Badawi, A., Bates, J., Bergamaschi, F., Cousins, D. B., Erabelli, S., Genise, N., Halevi, S., et al.

  3. IEEE TETC2021

    Towards the AlexNet Moment for Homomorphic Encryption: HCNN, the First Homomorphic CNN on Encrypted Data with GPUs

    Al Badawi, A., et al.

  4. IACR TCHES2018

    High-Performance FV Somewhat Homomorphic Encryption on GPUs: An Implementation using CUDA

    Al Badawi, A., Veeravalli, B., Mun, C. F., & Aung, K. M. M.

All publications and patents

At a glance

I run research on fully homomorphic encryption and privacy-preserving AI, with a focus on what it takes to deploy them at production scale. My contributions include PALISADE (now OpenFHE), a co-authored open-source FHE library now used across industry and academia; the first GPU homomorphic CNN (HCNN); the RNS variants of BFV that became standard building blocks in modern FHE libraries; and the technical leadership of DPRIVE, a $15M DARPA-funded effort that delivered a 1 GHz, 12 nm FHE accelerator from concept to tape-out-ready RTL.

I hold a Ph.D. in Electrical and Computer Engineering from the National University of Singapore. I have built and led research teams in industry, government, and academic settings, and I currently work on making large language model inference run under FHE with practical latency.

Open to principal/staff Cryptography & ML-Systems roles. I am also open to research collaboration, technical advising for founders building privacy-first infrastructure, and program committee service in cryptography and trustworthy ML. Feel free to reach out via my contact page.

For the agenda behind the work, see my research statement. For how it is used in practice, see my research impact. For shipped open-source projects, see code.

Also