<

Irem Ergun

I am a Ph.D. candidate at University of California, Riverside. I work on developing efficient, privacy-preserving distributed machine learning algorithms.

I am currently in the job market. I am interested in roles such as machine learning engineer, research engineer (ML), data scientist, and software engineer.

About Me

  • I am a Ph.D. candidate at University of California, Riverside CSE department.

  • My research interests include machine learning, federated learning, privacy and fairness.

  • Prior to starting my Ph.D., I have worked in the industry for some time. I have around 3 years of software engineering and system design experience.

  • I have worked on several natural language processing and computer vision projects.

  • I am knowledgeable in how to develop, orchestrate and maintain machine learning workflows. (#MLOps)

  • I am looking for machine learning engineer, research engineer (ML), data scientist, and software engineering roles.

  • I am passionate about creating state-of-the-art machine learning systems and I have multiple references that can vouch for me.

Interested in hiring me? Send me an email or reach out to me via LinkedIn!


Research:

  • Sparsified Secure Aggregation: Leverages gradient sparsification for communication efficient federated learning with provable privacy and convergence guarantees. Achieves up to 7.8x improvement in communication while also reducing the wall-clock training time, compared to benchmarks.

  • Secure Aggregation for Privacy-Aware Federated Learning with Limited Resources: Proposes a lightweight version of the secure aggregation protocol for private federated learning with limited resources. Reduces the communication overhead per round up to 10x without sacrificing significantly from model performance and privacy. Accepted at ICLR'22 Socially Responsible ML Workshop.

  • Sparsified secure aggregation with tunable privacy and performance guarantees. (Ongoing)

  • Optimal split-state non-malleable codes: Worked on developing constant rate, split state non-malleable codes using polynomials as my advancement to candidacy project.

  • Developed semi-supervised learning methods for breast cancer type detection problem using genomics data. The project was selected as the best undergraduate research project of the semester.

  • Check out my Github page for my other projects!
federated learning

Teaching:

  • (UCR) CS141:Algorithms-> Taught two lab sessions per week, held office hours, and graded exams.

  • (UCR) CS6:Effective use of the World-wide Web-> Held office hours, answered questions about HTML/CSS, and graded labs.

  • (Bilkent Uni) CS102:Intro to Algorithms, CS114:Intro to Programming with Java-> Helped students during Java lab sessions and graded labs.

  • (Delta Analytics) Global Teaching Fellow->Learned pedagogical concepts about teaching machine learning concepts all around the world. My graduation lecture about understanding microaggressions in Turkish using NLP can be found here.

  • Held a workshop in Turkish about federated learning.

  • Held a workshop in Turkish about reading academic papers to an audience of over 70 people.
  • I would love to chat, especially about prospective job opportunities, research, and music! Feel free to reach out to me!

  • Email: iergu001 AT ucr.edu

  • I can usually be found around Winston Chung Hall.