Ayushi Agarwal

PhD Student, Prime Minister Research Fellow
MARG, IIT Delhi Computer Science

About Me

I am a PhD Student (since 2019) and Prime Minister Research Fellow (since 2020) in Amar Nath and Shashi Khosla School of Information Technology (SIT) at Indian Institute of Technology Delhi. I work with the Memory and embedded Architecture Research Group (MARG) headed by Prof. Preeti Ranjan Panda. I have done my Bachelors in Electronics and Communication Engineering from National Institute of Technology Allahabad. I have worked as a Design Engineer in Qualcomm Design Center, Bangalore for 3 years. Before joining PhD, I have worked as a Research Engineer in the Green IC Group led by Prof. Massimo Alioto in National University of Singapore for a year.

My broad interests lie in the field of Heterogeneous System Architecture, System design and hardware for Machine Learning. Specifically my current research focus is on Heterogeneous System-On-Chip design with Machine Learning Accelerators.

I spend my spare time reading books, listening to soft music, crafting and coloring. I am also a volunteer for NGO that goes by the name GirlsWhoCode (Globally) and SheCodesIndia. I also love traveling and meeting new people.

Experience and Education

Timeline

My latest CV can be accessed using this link: CV


  • Delhi

    July 2019-Now

    PhD Student, PMRF Fellow, IIT Delhi

    Department: Amar Nath and Shashi Khosla School of Information Technology
    Research Interests: Computer Architecture, System-on-Chip, AI Accelerators
    Advisor: Preeti Ranjan Panda


  • Bangalore

    March 2021-July 2021

    Technical Intern, NXP Semiconductors

    Performance Architecture Group, MCU/MPU Engineering


  • Delhi

    July 2018-Dec 2018

    Project Engineer, IIT Delhi

    Project: Request Arbitration in Last-Level Caches
    Intelligent bandwidth partitioning at Last-Level Cache
    Paper


  • Singapore

    July 2017-June 2018

    Research Engineer, National University of Singapore

    Green IC Group
    Deep Learning Architectures


  • Bangalore

    June 2014-June 2017

    Design Engineer, Qualcomm Design Center

    SoC Front End Design Group
    RTL design and Implementation of Network-On-Chip for Qualcomm MSM and MDM devices


  • Bangalore

    Summer 2013

    Qualcomm Internship

    Automatic Validation of NoC RTL-Performance Analysis


  • Allahabad

    2010-2014

    B.Tech, National Institute of Technology, Allahabad

    Bachelors in Electronics and Communication Engineering
    CGPA: 9.65/10

Research

My Ongoing Research

I am currently working on Hardware for Machine Learing running on Edge Devices. I am also working on Heterogeneous SoC design with domain-specific accelerators. Custom hardware-based neural network accelerators have been known to surpass their general-purpose processor equivalents in terms of both throughput and energy-efficiency. On-chip inference on edge devices for IOT and other AI applications has also gained traction. I am looking into application-tailored accelerators co-designed with efficient network training models and their efficient integration on edge-devices.

Target System Architecture

The figure depicts a system with multiple clusters of processor cores with private L1 and dual-core shared L2 caches. All the processors share a last-level cache. The last-level cache is usually banked and is connected to the DRAM controller by a system interconnect. These kind of architectures are commonly used in edge devices. Our work focuses on system-level issues occuring because of running machine learning applications on these systems. These applications are compute and memory intensive and are extremely demanding of the shared resources of the system.

Seldom the performance of other applications running on the system is affected at the cost of high performance requirement of these applications. Since more and more of these models are being run on edge devices, there is a need to understand the system-level issues mostly at the shared system resources to improve the complete system performance instead of focussing on application centered performance.

Publications

1. Sakshi Tiwari, Shreshth Tuli, Isaar Ahmad, Ayushi Agarwal, Preeti Ranjan Panda, Sreenivas Subramoney: REAL: REquest Arbitration in Last Level Caches , ACM Transactions on Embedded Computing Systems (TECS) 18(6) 1-24.

Coursework

Coursework done during PhD

Year: 2019-2020

First Semester:
  • COL 718: Architecture of High-Performance Computers ; Course Webpage
    Taught By : Prof. Sorav Bansal
    Assignments and Labs : Github Link
  • COL 730: Parallel Programming
    Taught By : Prof. Kolin Paul
    Assignments and Labs : Parallel Matrix-Vector Multiplication, Parallel Edit Distance

  • Second Semester:
  • COL 774: Machine Learning
    Taught By : Prof. Parag Singla
    Assignments: Github Link
  • COL 869: Special Topics on Memory Concurrency
    Taught By: Prof. Soham Chakraborty
  • COL 812: System Level Design and Modeling
    Taught By: Prof. Anshul Kumar
  • HSL 800: Research Writing
    Taught By: Bijoy Philip

  • Year: 2020-2021

    Third Semester:
  • COL 861: Special Topics in Hardware Systems (Hardware and Machine Learning)
    Taught By : Prof. Preeti Ranjan Panda
  • Professional Activities

    Teaching Assistantship, Services, Awards and Talks

    Teaching Assistanship

    Role as a Teaching Assistantship includes defining the lab exercises, assisting students with doubts and lab work and assistance provided in evaluation.

    Year Semester 1 Semester 2
    2019-2020 COL 719: Synthesis of Digital Systems COL 216: Computer Architecture
    Taught By: Prof. Preeti Ranjan Panda Taught By: Prof. Preeti Ranjan Panda
    2020-2021 COL 215: Digital Logic and System Design TA off
    Taught By: Prof. Anshul Kumar
    2021-2022 COL 215: Digital Logic and System Design COL 216: Computer Architecture
    Taught By: Prof. Vireshwar Kumar
    and Prof. M Balakrishnan
    Taught By: Prof. Anshul Kumar
    2022-2023 COL 215: Digital Logic and System Design COL 719: Synthesis of Digital Systems
    Taught By: Prof. Preeti Ranjan Panda Taught By: Prof. Preeti Ranjan Panda

    Services: Peer-Reviews

    IEEE Journal on Emerging and Selected Topics in Circuits and Systems (IEEE JETCAS)
    Year: 2018, 2019

    IEEE Embedded System Letters (IEEE ESL)
    Year: 2020, 2021, 2022, 2023

    Design, Automation and Test in Europe Conference (DATE)
    Year: 2021

    Design Automation Conference (DAC)
    Year: 2021, 2022

    Asia and South Pacific Design Automation Conference (ASP-DAC)
    Year: 2022

    VLSI Design Conference (VLSID)
    Year: 2023

    ACM Transactions on Design Automation of Electronic Systems (TODAES)
    Year: 2023

    International Conference on Compilers, Architectures, and Synthesis for Embedded Systems (Embedded Systems Week: CASES)
    Year: 2023


    Awards

    Prime Minister Research Fellowship, May 2020. (PMRF Webpage)
    Selected for receiving the prestigious Prime Minister Research Fellowship offered by Ministry of Human Resource Development (MHRD), Government of India, from May 2020

    Cadence Tensilica Hackathon at VLSI Design Conference 2019. (Pictures)
    Won the first prize in the Cadence Tensilica Hackathon with team-mates Anupam Sobti and Saurabh Tewari

    Institute Gold Medal 2014
    Awarded the Institute Gold Medal for being the institute topper in Electronics and Communication Engineering, 2014

    Late (Dr.) Malay Raj Mukherjee Gold Medal 2014
    Awarded the Late (Dr.) Malay Raj Mukherjee Gold Medal for standing first amongst the students of B.Tech (Electrical Engineering), B.Tech (Computer Science and Engineering) and B.Tech (Electronics and Communication Engineering)

    OPJEMS scholarship 2012
    Awarded the O.P. Jindal Engineering and Management scholarship by Jindal Steel and Power Ltd. for exhibiting outstanding competency, leadership and communication skills.

    My Personal Blog

    I like to read books when I am not working and I write and share my journey as a female scientist/engineer.
    Disclaimer: The Views in this page are my own.
    The institute does not endorse or represent any of them.

    Books

    Book that I have read and would like to share

    MERI AWAAZ HE PEHCHAAN HAI

    People will forget my name but my voice will echo long after I perish!

    Volunteering

    "Be a rainbow in someone else's cloud" - Heartfulness

    Contact Me

    307 VLSI Design Lab, 3rd Floor SIT,
    IIT Delhi, New Delhi 110016 (India)