Logo

Haseeb Ashfaq

CV

GitHub

Tech Blog

LinkedIn

Google Scholar

My legal name is Muhammad Haseeb.

Ph.D. Computer Science candidate at New York University
Working in Systems group Systems@NYU
Office: Room 444, 60 5th Avenue, New York, NY 10011
mh6218@nyu.edu
haseeb.luminite@gmail.com

Research Interests

  1. Computer Networks
  2. Distributed Systems
  3. Cloud Computing
  4. AR/VR streaming

Research Papers

Network Support For Scalable and High Performance Cloud Exchanges

  • ACM SIGCOMM 2025
  • Haseeb Ashfaq, Jinkun Geng, Daniel D-Cavalcanti, Xiyu, Hao, Ulysses Butler, Radhika Mittal, Srinivas Narayana, Anirudh Sivaraman
  • Cited by researchers and engineers from Jane Street and Google.

ParserHawk: Hardware-aware parser generator using program synthesis

  • ACM SIGCOMM 2025
  • Xiangyu Gao, Jiaqi Gao, Karan Kumar, Haseeb Ashfaq, Enan Zhai, Bili Dong, Joseph Tassarotti, Srinivas Narayana, Anirudh Sivaraman

Towards Efficient Transmission of 3D Point Clouds Via Adaptive Encoding and QUIC

QuEST: Fast, Expressive, and Cheap Analytics for Distributed Traces Using Cloud Storage

  • CloudDB Workshop @ VLDB'23
  • Authors: Jessica Berg, Haseeb Ashfaq, Haiming Chen, Yaojia Ju, Anirudh Sivaraman, Ravi Netravali and Srinivas Narayana

To Block or Not To Block: Accelerating Mobile Webpages On-The-Fly Through JavaScript Classification

  • ICTD'22
  • Authors: M Chaqfeh, M Haseeb, W Hashmi, P Inshuti, M Ramesh, M Varvello, F Zaffar, L Subramanian, Y Zaki

Using Application Layer Banner Data to Automatically Identify IoT Devices

Ongoing Research Projects

High Performance and Scalable Multicast As A Cloud Service

  • Designed and implemented a low-latency and scalable multicast service
  • Used kernel-bypass techniques (DPDK) for achieving ultra-low latency
  • Achieved better latency and scalability than the multicast provided by AWS Transit Gateway

Real Time Volumetric Videos streaming

  • Designed and implemented a scalable streaming system that can adapt to network conditions and serve heterogeneous clients using minimal resources
  • Designed an encoder/decoder for point cloud data that can tolerate packet losses
  • Desgined a low latency transcoding mechanism for point clouds

Service Mesh Based Microservices Authorization For Byzantine Environments

  • Designed a new access control system, Path Aware Access Control that takes into account the path a user request follows in a microservices application before allowing it to access some microservice(s)

Experience

Software Engineering Intern (PhD) at Google, Sunnyvale CA, USA

  • Part of AI and Infrastructure organization, worked on improving robustness of ML infrastructure
  • Developed a tool, hstprof, for profiling GPU/TPU workloads using high speed network telemetry
  • hstprof enabled fine grained view of the network traffic of ML jobs, reporting utilization at 100 microseconds timescale
  • Used hstprof to analyze Gemini training clusters to investigate several regressions including packet drops and high latencies
  • Found imbalance of packet queues across memory banks of switches which caused performance degradation

Networking Research Intern at Nokia Bell Labs, Murray Hill, NJ, USA

  • Led a project for designing and developing a streaming service for AR/VR content
  • Designed and implemented a scalable streaming system that can adapt to network conditions and serve heterogeneous clients using minimal resources
  • Designed an encoder/decoder for point cloud data that can tolerate packet losses

Software Engineer at PosterMyWall, Pakistan

  • Secured the product website by eliminating critical security vulnerabilities (XSS, CSRF, IDOR, potential Brute Force Attacks)
  • Designed and implemented an access control system for different tools of the company
  • Setup CI/CD pipeline along with a testing infrastructure
  • Created a disaster recovery mechanism for the company’s infrastructure hosted on AWS
  • Worked with various AWS services: EC2, S3, CloudFormation, CloudWatch, CodeDeploy

Patents

A METHOD TO ENABLE FAST TRANSMISSION AND PROCESSING OF 3D TELEPRESENCE DATA ENCODED AS OCTREES

  • Approved by Nokia's Board "Wireline Networks Patent Board".
  • Under submission to USPTO.

Awards

Nokia Bell Labs Outstanding Student Research Award 2023

  • For my work on volumetric videos streaming.
  • Check out the wall plaque Bell Labs sent me.