Ahmet Fatih Inci portrait

Ahmet Inci

Ph.D.
Electrical and Computer Engineering
Carnegie Mellon University

Email: inciaf [AT] gmail [DOT] com

Affiliation: [EnyAC] [OPAL]
  • I am a Senior Deep Learning Performance Architect at NVIDIA, working on future hardware architectures and optimizations to advance the state-of-the-art in deep learning performance and energy-efficiency. Previously, I was a Machine Learning Engineer in Apple Neural Engine Compiler Team at Apple. I received my Ph.D. from CMU, co-advised by Prof. Diana Marculescu and Prof. Gauri Joshi. My dissertation was titled "Scalable and Efficient Systems for Deep Learning". Before joining CMU, I received my B.Sc. degree in Electronics Engineering at Sabanci University.

    My research interests include Systems for ML, HW/SW Co-Design, and Efficient Deep Learning.

    My Ph.D. research focused on designing scalable and efficient systems and ML models using HW/ML model co-design techniques to achieve the best of both worlds. I worked on quantization-aware DNN accelerator and model co-exploration through architecture-level modeling and efficient design space exploration. Before that, I worked on scalable and efficient reinforcement learning training on CPU-GPU systems. Additionally, my previous work has explored how to utilize emerging non-volatile memories in GPU architectures for DL workloads.

    Publications

    Work Experience

    NVIDIA - Deep Learning Compute Architecture
    January 2024 - Present




    Apple - Apple Neural Engine Compiler Team
    August 2022 - January 2024
    Research and development on neural engine compiler for ultra-low power devices



    NVIDIA Research - Architecture Research Group in collobaration with AI Research
    May 2021 - August 2021
    Optimizing Power Management of Deep Learning Systems with Reinforcement Learning



    NVIDIA Research - Architecture Research Group in collobaration with AI Research
    May 2020 - August 2020
    Towards Scalable and Efficient Reinforcement Learning on CPU-GPU Systems


    ARM - ML Technology Group
    May 2019 - August 2019
    NASH: Neural Architecture Search for Heterogeneous Systems


    Cadence Design Systems - Virtuoso ML Team
    May 2018 - August 2018
    ML-based Recommendation System for EDA Tools




    Service