Career Overview

Work Experience

Google

Software Engineer III (L4), Google Cloud

Seattle, WA, USA Manager: Vicky Xu
Sep 2024 — Present
  • Bare Metal Cloud Infrastructure: I engineer Google Cloud Platform (GCP) solutions to develop, maintain, and operationalize Bare Metal cloud solutions for demanding applications. The Google Computing Engine (GCE) Node Bare Metal servers allow users to run specialized high-performance computing (HPC) workloads, e.g., third-party virtualization software, applications with low-level access to the server, applications with ultra-low-latency (ULL) networking requirements, computationally demanding AI workloads, etc. Our team manages the development and availability of regionally distributed cluster of servers with high-performance connection with a low-latency network fabric.
    • I was involved in testing through the CI/CD pipeline and overall DevOps for building several Bare Metal shapes.
    • I worked on the first arm-based bare metal shape Google Cloud developed in 2025-2026. Here is a Google Cloud Blog as the public announcement of the product C4A Metal.
    • I worked on the first GPU-based bare metal Google Cloud developed which went GA around the end of Q1 2026, A4X Max Metal.
  • AI Tooling: I made a small sub-team of 4 engineers and led the design and development of an end-to-end integrated agentic AI tool to resolve issues for a tier-2 oncall (with SLO of 30 minutes) for GCE Bare Metal team. The tool utilizes a Retrieval Augmented Generation (RAG) agent, a vector database, and integrates with Google's internal Large Language Model (LLM) and provides a Frontend to the user. This tool increased the Bare Metal team's productivity to an extent that it motivated a number of other teams in the Org to leverage from the tool and integrate their knowledge corpus.
  • Observability: I develop, maintain, and own a number of internal dashboards to maintain health and monitor capacity of Bare Metal fleet. This improves the decision making for the team on a daily basis.
  • Hackathon: I led a small sub-team of 4 engineers for a hackathon entry for AI2 hackathon with an early version of the oncall AI tool.
  • Interviews: I took ~50 interviews of Software Engineering Full-time, SWE Intern, and Research Intern L3 - L5 interview candidates.
  • Volunteer: I took ~20 Google Champion calls (mock interviews).

Software Engineering Ph.D. Intern, YouTube

San Bruno, CA, USA Hosts: Dr. Jie (Jeremy) Zhou and Dr. Kevin Chen
May 2023 — Aug 2023
  • UVQ for YouTube Shorts: I collaborated closely with the playback experience, media algorithms, transcoding, and edge streaming teams and extended UVQ to YouTube Shorts. As part of this work, I evaluated multiple ML models through A/B testing to comprehend subjective perception of video quality and learn from contextual factors (such as content type, device type, geolocation, and other streaming objectives). Here is a Google Research Blog that introduces the UVQ model.
  • Intern Learning: Personal Branding, Fundamentals of Innovation, Impactful Performance Writing.
  • Volunteer: I helped job seekers by reviewing their resume.

Stony Brook University

Machine Learning Intern, Strategic Partnership for Industrial Resurgence (SPIR)

Stony Brook, NY, USA Hosts: Prof. Dantong Yu and Yu Sun
Jan 2024 — Aug 2024
  • Explainable Graph Learning: I collaborated with researchers from Brookhaven and Los Alamos national labs and designed an explainable graph neural network (GNN) to detect heavy flavor decays in real-time for the sPHENIX project, a high-energy nuclear physics experiment at Brookhaven National Laboratory.
  • Terabits data transfer: I worked with engineers at Sunrise Technology Inc. incubated at the Center of Excellence Wireless and Information Technology (CEWIT) on the development of a low-power embedded system (NVIDIA Jetson devices serving as DPUs) enabling high-speed data transfer (100 Gbps) across distributed file system nodes (CephFS) over the internet using RDMA protocol RoCEv2, optimizing performance for high-performance computing applications (funded by and in collaboration with the Department of Energy). Feel free to check out this demo.

Research Project Assistant

Stony Brook, NY, USA Advisor: Prof. Michael A. Bender
Jun 2019 — Dec 2022
  • Cache-efficient algorithms: I designed an empirical framework to evaluate when programs are cache- and memory-adaptive for external memory applications. These programs maintain good performance despite memory fluctuations, which are typically found in multicore, multithreaded, shared-memory, and cloud systems. I provided an algorithm design principle and reference implementation for cache-adaptive algorithms to solve large external-memory problem instances for fundamental problems such as cubic-time matrix multiplication, sorting, and dynamic programming algorithm for the longest common subsequence (LCS).
    • Check out our paper in 30th Annual European Symposium on Algorithms (ESA 2022)
    • I extended this work and built a framework to evaluate the performance of external-memory algorithms with multi-threading in a multi-program HPC environment and broadly offers a recipe for algorithm design for multi-threading ecosystem. Check out the code.
  • Machine-learning augmentation: I redesigned traditional online algorithms rent-or-buy problems with augmentation by single and multiple ML oracles accounting for arbitrarily fluctuating discounts on the rent of the resource. In the subsequent work, I revisited conventional online decision-making problems, such as list access problem, job scheduling problem, secretary selection problems, and green and parallel paging problem.
  • Filesystem aging: I collaborated with a large group of researchers, engineers, and professors to evaluate the degradation of read performance across five production filesystems, ext4, btrfs, xfs, zfs, and f2fs, owing to poor data layout and disk fullness. This evaluation, conducted using microbenchmarks and application-level fragmentation benchmarks, demonstrated that the degradation could be mitigated with a $B^{\varepsilon}$-tree-based write-optimized in-kernel filesystem like BetrFS.
    • Check out our article on ArXiv.

Mentor, Women in Science and Engineering (HS-WISE)

Stony Brook, NY, USA Supervisor: Jacquelyn Gatta
Sep 2019 — Apr 2023
  • Lessons: Basics of Probability and Thinking Algorithmically.
  • I mentored high-school students across Long Island, NY and encouraged them to pursue a career in science and engineering.
    • Shoreham-Wading River High School (2019-20).
    • Comsewogue High School (2020-21).
    • Hauppauge High School (2021-22).
    • Bellport High School (2022-23).
  • Check out the recommendation letter I received from the Program Director of WISE at Stony Brook, Jacquelyn Gatta.

Graduate Teaching Assistant

Stony Brook, NY, USA
Sep 2018 — May 2919 & Jan 2023 — Apr 2023
  • Graded papers and homework assignments, proctored exams, and took weekly office hours.
    • Theory of Database Systems (CSE 532) in Spring 2023 under Prof. Fusheng Wang.
    • System Fundamentals - II (CSE 320) for Spring 2019 under Prof. Eugene Stark.
    • Fundamentals of Information Technology (ISE 218) in Fall 2018 under Prof. Kevin McDonnell.

Visiting Research Student Intern

Lawrence Berkeley National Laboratory | Berkeley, CA, USA

Sep 2023 — Dec 2023
  • Dynamic GAP: I Expanded the experimental framework of the GAP@Berkeley project to incorporate dynamic graph algorithms and designed an experimental framework to study the performance of external-memory algorithms (APIs) for large sparse dynamic graphs. I studied batch-parallel algorithms for processing large sparse time-evolving graphs and conducted a comparative study with the top node-wise parallel static algorithm using this framework. Check out the GitHub repo for this work.

Cloud and Networking Intern

Nokia Bell Labs | Murray Hill, NJ, USA

Hosts: Dr. Edward Grinshpun and Chuck Payette
Jun 2022 — Aug 2022
  • Augmenting L4S with predictions: I researched out-of-band ML-based prediction-enhanced congestion control algorithms tailored for low-latency, high-volume, variable-bitrate applications, such as live video streaming, within 5G wireless network systems.
  • Intern Learning: I completed Harvard ManageMentor Leading People by Harvard Business Publishing and End-to-End 5G Networking by Nokia Bell Labs.

Research Engineer

National University of Singapore | Singapore

Manager: Prof. Dipti Srinivasan
May 2018 — Aug 2018
  • I reviewed the performance of Multi Objective Evolutionary Algorithm based on Decomposition (MOEAD) techniques experimentally using Matlab.

Consultant, Technology Consulting

Pricewaterhouse Coopers India | Kolkata, WB, India

Manager: Sudipto Sarkar
Jul 2016 — Sep 2017
  • I was involved in onsite visits for requirement gathering, and engaging with client for system co-design. I co-led the technical development of an ERP solution. The system aimed to streamline buisness processes and generate reports for finance, procurement, and inventory management for an international client in the power generation and distribution sector.
  • I built a payroll automation system using DotNet technologies using MVC architecture.
  • Volunteer activities: I took active roles of speaker and reviewer at the Toastmasters International Club at PwC India.

Undergraduate Research Assistant

Jadavpur University | Kolkata, WB, India

Advisor: Prof. Debangshu Dey
Jun 2015 — May 2016
  • I researched on image and video processing and designed algorithms for affective computing and programmed a methodology to detect emotion. We studied Bi-dimensional Empirical Mode Decomposition (BEMD) based feature extraction, Principle Component Analysis (PCA), Linear Discriminant Analysis (LDA) based dimensionality reduction, Gray-level Co-occurrence Matrix (GLCM), Histogram of Oriented Gradients (HOG), Local Ternary Pattern (LTP) based feature elimination, and Multi-class Support Vector Machine (MSVM), k-Nearest Neighbor (kNN) based classification. We did the empirical evaluation on 3 datasets (JAFFE, Cohn–Kanade, and eNTERFACE) using Matlab.
  • I designed algorithms for biomedical image processing and cancer detection using optical colonoscopy videos by cross-wavelet feature extraction.

Undergraduate Summer Intern

IIT Kharagpur | Kharagpur, WB, India

Advisor: Prof. Jayanta Mukhopadhyay
Jun 2014 — Aug 2014
  • I designed a portable electronic device for non-invasive continuous measurement of blood pressure.

Hackathons