Dev

Curriculum vitae

This is the full CV, if you are looking for something shorter here is my résumé

Education

University of California Santa Cruz, Santa Cruz, CA, USA.

PhD. Computer Science (June ‘24)
Dissertation: Enhancing Flash Storage Performance and Lifetime with Host-Guided Data Placement.
MS Computer Science (2022)

Savitribai Phule Pune University, Pune, India.

BE Computer Engineering (2016)
First Class With Distinction

Professional

Software Engineer, Azure Storage ◆ Microsoft Corp.

April 2024–present
Working on scale-out block storage for Azure on pipelined Data Processing Units (DPUs) (C, C++).

Architecture Enabling Intern, SOLAB ◆ SK Hynix America Inc.

June–Sept 2019
Worked on two projects:
(i) Analysis an simulation of a variety cold storage workloads for QLC flash in Python, and
(ii) Reduce Ceph’s recovery bandwidth with inline compression (C++).
Awards: Most Valued Professional, September 2019.

Engineering Intern ◆ Riverbed Inc.

June–Sept 2017
Refactored and modernized outdated network statistics collection codebase to implement a multi- threaded, non-blocking solution in golang.

Center for Research in Storage and Systems, UC Santa Cruz ◆ Graduate Student Researcher

(i) Setup and maintenance of research clursters: ldap, NFS, provisioning, troubleshooting, sysadmin, and networking.
(ii) Mentoring, collaboration, feedback, and assitance to other CRSS research projects.
(iii) Teaching Assistant for courses in operating systems, databases, and programming.

Research

Shimmer: Hint Generation and Data Placement on Modern SSDs

Shimmer is a dynamic library written in Rust which uses a shim layer to intecept libc calls and redirect them with workload hints. Shimmer improves write throughput by 30–90% on ZNS SSDs while reducing tail latency by 14× for RocksDB, MongoDB, and Cachelib.

Persimmon: A filesystem for Zoned Namespaces

Persimmon updates F2FS metadata to be append-only, optimizing the filesystem for the ZNS interface, reducing garbage collection, tail latency, and improving storage utilization. Technologies: C, filesytsems, and kernel programming.

Analysis of peta-scale scientific archives

Analysis of a year of petabyte filesystem activity at CERN, spanning 2.4 billion file events done in Apache spark, with suggestions for optimizations, and comparison with historical trends.

Publications

Teaching

Teaching Asistant for CMPS 10, 11, 12A (Introductory Programming Courses), CSE 13S (Systems Programming and C), CSE 134 (Embedded Operating Systems), CSE 180, 182 (Introduction to Data Management Systems).
Quarters: Spring 17—Spring 18, Spring 21, Fall 21, Fall 22, Winter 22, Fall 23, Winter 23.

Invited Talks

Shimmer: Interposition on modern SSDs

  • Google: Android Block Storage, F2FS team, August 25, 2023.
  • George Washington University, Systems group, November 8, 2023
  • Languages, Systems, and Data Seminar, UC Santa Cruz, January 26, 2024
  • Nutanix Inc. February 21, 2024

Persimmon: an append-only ZNS-first filesystem

  • The 41st IEEE International Conference on Computer Design, Washington DC, November 7, 2023

Pre-Conference Seminar: SSD Controllers

  • Flash Memory Summit 2023, August 7, 2023.

Append is Near: Log-based Data Management on ZNS SSDs.

Trace Analysis of Large Scale Storage Systems.

Service

  • Organizer: !!Con West— a conference on the joy, excitement, and surprise of computing. 2019, 2020.
  • Administrator: discuss.systems — a Mastodon instance for systems researchers managed by  Dan, Irene, and me.