Experience

Research Experience

• Achieveing Fairness and Performance while Serving LLMs Amid Diverse Applications. Summer 2024
  Research Intern, AIOps and Efficient AI at Microsoft
  Advisor: Dr. Ankur Mallick and Dr. Anjaly Parayil
  

  Developed an application-characteristic-aware request throttling mechanism combined with a weighted service counter-based scheduling technique to mitigate abusive behaviors and promote fairness in Microsoft's CoPilot LLM serving platform, improving the experience for millions of users.

• Workload characteristic-aware ML Systems. Fall 2020 - Present
  Ph.D. Thesis Research, Advisor: Dr. Ali R Butt
  

  Existing platforms frequently waste resources when handling DDL, FL, and LLM workloads, as they lack optimization for the unique demands of these tasks. My thesis aims to infuse intelligence into these systems to maximize quality, efficiency, autonomy, and adaptability with minimal human intervention. Additionally, it focuses on enriching user experience and satisfaction in cloud services through AI-driven insights, while integrating AI/ML across the software development lifecycle to boost productivity and streamline processes end-to-end.

• Container Usage Patterns in HPC Scenarios. Fall 2019 - Summer 2020
  PhD Research Project, Advisor: Dr. Ali R Butt
  

  Although the usage of containers has peaked in the recent years, its characteristics when used in HPC scenarios had been left unexplored. The goal of this project was to understand those patterns so that bottlenecks could be removed and improved architectures could be designed.

• Autotuning a High-Performance Datatype Engine. Summer 2021
  Intern, Mathematics and Computer Science(MCS) at Argonne National Lab
  Advisor: Dr. Yanfei Guo and Dr. Shintaro Iwasaki
  

  Examined current performance with different datatypes in Yaksa Pack/Unpack Library of MPICH, developed scripts to implement optimizations in performance, created benchmarks for providing insights into performance optimization scenarios, and designed an auto-tuner to explore a search space for optimizations.

• Approaches for Improving Graph Coloring in Large Networks. Fall 2018 - Spring 2019
  UG Research Project, Advisor: Dr. Erdem Sariyuce
  

  Explored on improving graph coloring in massive graph networks and identified methods for analyzing large and distributed stock market datasets.

• Power Performance Trade-offs for Mobile Devices in Next Gen. WiFi. Spring 2018
  UG Research Project, Advisor: Dr. Dimitrios Koutsonikolas
  

  Devised algorithms for understanding power consumption by mobile devices using different cores and packet transfer rates.

Teaching Experience

• CS2506 - Computer Organization II. Fall 2019 - Present
  Teaching Assistant, Virginia Tech
  Intermediate course in Computing for CS majors.
  Responsibilities: Teaching, Creating Course content, Setting lab servers, TA coordination
  Instructor: Dr. Xun Jian. Course enrollment: ~350 students.
• CS321 - Real Time Embedded Systems. Fall 2018
  Teaching Assistant, University at Buffalo
  Advanced course in computing for CS majors.
  Responsibilities: Taking office/recitation sessions, Creating course content, Grading, Exam proctoring
  Instructor: Matt Stock. Course enrollment: ~200 students.
• CS487/587 - Data Intensive Computing. Spring 2019
  Teaching Assistant, University at Buffalo
  Advanced course in computing for CS majors and graduate students.
  Responsibilities: Taking office/recitation sessions, Grading, Exam proctoring
  Instructor: Dr. Bina Ramamurthy. Course enrollment: ~200 students.