deeksha_dangwal_headshot
deeksha [at] ucsb [dot] edu

About me

I am a PhD candidate in Computer Science at UC Santa Barbara. I study computer architecture with Prof. Tim Sherwood and I am interested in the design of private computer systems and applications. My interdisciplinary research spans computer architecture, privacy, computer vision, and machine learning.

In my PhD, I have explored the privacy of program traces where the intent is to minimize information leakage to ensure safer sharing of program behavior for co-optimization. The key trade-off is balancing information leakage while maintaining utility of the privatized traces. Trace wringing was published in ASPLOS 2019 and recognized in IEEE Micro’s Top Picks from Computer Architecture Conferences 2020. Previously, I have worked on PyRTL, a Python-based RTL specification language, and built the OpenTPU and cryptographic cores on it. I have also worked on Charm, a high-level architecture modeling language.

During my internship at Facebook Reality Labs Research, I worked on building privacy into AR/VR pipelines. With a novel reverse engineering attack, I demonstrated that feature descriptors, previously considered safe to share, can be used to reconstruct raw user images. I present mitigations that trade off privacy (quality of reconstruction) and utility (in terms of localization accuracy). At Microsoft Research, I developed an automated framework for resource- and architecture-aware graph partitioning and fine tuning quantized neural network layers on the Brainwave NPU.

Recent News

I am on the job market. Please reach out to me if you are hiring.

Education

2016-present
Ph.D. in Computer Science, UC Santa Barbara
2014-2016
M.S. in Electrical and Computer Engineering, UC Santa Barbara
2010-2014
B.E. in Electronics and Instrumentation, M. S. Ramaiah Institute of Technology

Experience

Oct 2020-Jan 2021
Part-Time Student Researcher, Facebook Reality Labs Research, Seattle WA
June 2020-Oct 2020
Research Intern, Facebook Reality Labs Research, Seattle WA
June 2018-Sep 2018
Research Intern, Microsoft Research, Seattle WA
June 2016-Sep 2016
Research Intern, Oracle Labs, Austin TX

Awards

March 2021
UCSB Grad Slam Runner-up
November 2020
Rising Stars in EECS 2020, UC Berkeley
June 2020
IEEE Micro Top Pick: Trace Wringing for Program Trace Privacy
June 2020
Fiona and Michael Goodchild Graduate Mentoring Award
June 2020
Outstanding Graduate Student Award

Conference

Porcupine: A Synthesizing Compiler for Vectorized Homomorphic Encryption
Meghan Cowan, Deeksha Dangwal, Armin Alaghi, Caroline Trippel, Vincent T. Lee, and Brandon Reagen

Proceedings of the 42nd ACM SIGPLAN International Conference on Programming Language Design and Implementation (PLDI) June 2021

PDF
Safer Program Behavior Sharing through Trace Wringing
Deeksha Dangwal, Weilong Cui, Joseph McMahan, and Timothy Sherwood

Proceedings of the 24th International Conference on Architectural Support for Programming Languages and Operating Systems (ASPLOS) April 2019. Providence, RI

PDF Project Video Cite
Charm: A Language for Closed-form High-level Architecture Modeling
Weilong Cui, Yongshan Ding, Deeksha Dangwal, Adam Holmes, Joseph McMahan, Ali JavadiAbhari, Georgios Tzimpragos, Frederic T. Chong, and Timothy Sherwood

Proceedings of the 45th Annual ACM/IEEE International Symposium on Computer Architecture (ISCA) June 2018. Los Angeles, CA

PDF Project Video Code Cite
A Pythonic Approach for Rapid Hardware Prototyping and Instrumentation
John Clow, Georgios Tzimpragos, Deeksha Dangwal, Sammy Guo, Joseph McMahan, and Timothy Sherwood

Proceedings of the International Conference on Field Programmable Logic and Applications (FPL) September 2017. Ghent, Belgium

PDF Project Code Cite

Journal and Magazine Articles

Trace Wringing for Program Trace Privacy
Deeksha Dangwal, Weilong Cui, Joseph McMahan, and Timothy Sherwood

IEEE Micro: Micro's Top Picks from Computer Architecture Conferences, May-June 2020.

PDF Project Video Cite
Agile Hardware Development and Instrumentation With PyRTL
Deeksha Dangwal, Georgios Tzimpragos, and Timothy Sherwood

IEEE Micro 40, no. 4 (2020)

PDF Project Code Cite
Language Support for Navigating Architecture Design in Closed Form
Weilong Cui, Georgios Tzimpragos, Yu Tao, Joseph McMahan, Deeksha Dangwal, Nestan Tsiskaridze, George Michelogiannakis, Dilip Vasudevan, and Timothy Sherwood

ACM Journal on Emerging Technologies in Computing Systems (JETC) October 2019

PDF Project Video Code Cite

Preprints

Porcupine: A Synthesizing Compiler for Vectorized Homomorphic Encryption
Meghan Cowan, Deeksha Dangwal, Armin Alaghi, Caroline Trippel, Vincent T. Lee, and Brandon Reagen

arXiv, January 2021

PDF Cite

Workshops

SoK: Opportunities for Software-Hardware-Security Codesign for Next Generation Secure Computing
Deeksha Dangwal, Meghan Cowan, Armin Alaghi, Vincent Lee, Brandon Reagen, and Caroline Trippel

Hardware and Architectural Support for Security and Privacy (HASP) held in conjunction with 53rd International Symposium on Microarchitecture (MICRO), October 2020

PDF
PyRTLMatrix: an Object-Oriented Hardware Design Pattern for Prototyping ML Accelerators
Dawit Aboye, Dylan Kupsh, Maggi Lim, Jacqueline Mai, Deeksha Dangwal, Diba Mirza, and Timothy Sherwood

The 4th Workshop on Energy Efficient Machine Learning and Cognitive Computing for Embedded Applications (EMC2) held in conjunction with the 46th International Symposium on Computer Architecture (ISCA), June 2019, Phoenix, AZ.

PDF Cite
PyRTL in Early Undergraduate Research
Diba Mirza, Deeksha Dangwal, and Timothy Sherwood

Proceedings of the Workshop on Computer Architecture Education (WCAE) held in conjunction with the 46th International Symposium on Computer Architecture (ISCA), June 2019, Phoenix, AZ

PDF Cite

Patents

Deriving a concordant software neural network layer from a quantized firmware neural network layer
Jeremy Fowers, Daniel Lo, Deeksha Dangwal

Microsoft Technology Licensing LLC, US20200279153A1, September 2020

Cite