CV

PDF Version

Education

• Doctor of Philosophy - Electrical and Computer Engineering (4.00/4.00), 2024-Present
  Purdue University - The Elmore Family School of Electrical and Computer Engineering - Indiana, US.
  Advisor: Prof. Abolfazl Hashemi

• Master of Science - Electrical and Electronics Engineering (3.91/4.00), 2021-2024
  Bilkent University - Faculty of Engineering - Ankara, Turkey.
  Advisor: Prof. Süleyman Serdar Kozat

• Bachelor of Science - Electrical and Electronics Engineering (3.61/4.00), 2017-2021
  Middle East Technical University - Faculty of Engineering - Ankara, Turkey.

Research Background & Interests

• Optimization: constrained/unconstrained optimization theory, multi-objective optimization
• Trustworthy Deep Learning: robustness, privacy, memorization
• Distributionally Robust Optimization: invariant learning, subpopulation shift, domain generalization

Publications

Preprints

1. A. Fazla, E. Kaya, A. Upadhyay, A. Hashemi, “Lower Bounds and Proximally Anchored SGD for Non-Convex Minimization Under Unbounded Variance”, Under Review at NeurIPS 2026.
   Available: https://arxiv.org/abs/2604.16620

2. A. Upadhyay, A. Fazla, A. Hashemi, “Beyond Bounded Variance: Variance-Reduced Normalized Methods for Nonconvex Optimization under Blum-Gladyshev Noise”, Under Review at NeurIPS 2026.
   Available: https://arxiv.org/abs/2605.15314

Conference Papers

1. A. Fazla, A. Hashemi, “Mitigating Spurious Correlations with Memorization-Guided Dataset De-Biasing Selection”, Uncertainty in Artificial Intelligence, 2026.
   Available: https://arxiv.org/abs/2606.02830

2. K. G. Ince, A. Köksal, A. Fazla, A. A. Alatan, “Semi-Automatic Annotation for Object Detection”, Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV) Workshops, 2021, pp. 1233-1239.
   Available: https://doi.org/10.1109/ICCVW54120.2021.00143

Journal Papers

1. S. F. Tekin, A. Fazla, S. S. Kozat, “Numerical Weather Forecasting using Convolutional-LSTM with Attention and Context Matcher Mechanisms”, IEEE Transactions on Geoscience and Remote Sensing, 2024.
   Available: https://arxiv.org/abs/2102.00696
   Code: https://github.com/sftekin/spatio-temporal-weather-forecasting

2. A. Fazla, M. E. Aydin, S. S. Kozat, “Time-Aware and Context-Sensitive Ensemble Learning for Sequential Data”, IEEE Transactions on Artificial Intelligence, 2023.
   Available: https://doi.org/10.1109/TAI.2023.3319308
   Code: https://github.com/ardafazla/context-time-aware-ensemble

3. M. E. Aydin, A. Fazla, S. S. Kozat, “Hybrid State Space-based Learning for Sequential Data Prediction with Joint Optimization”, arXiv preprint arXiv:2309.10553, 2023.
   Available: https://arxiv.org/abs/2309.10553
   Code: https://github.com/mustafaaydn/lstm-sx

4. A. Fazla, M. E. Aydin, S. S. Kozat, “Joint Optimization of Linear and Nonlinear Models for Sequential Regression”, Digital Signal Processing, Elsevier, 2022.
   Available: https://doi.org/10.1016/j.dsp.2022.103802
   Code: https://github.com/ardafazla/jointoptimization

Professional Experiences

Purdue University, Research Assistant

Advisor: Prof. Abolfazl Hashemi
West Lafayette, IN

Stochastic Optimization, Aug 2025-Present

• Developed PASTA and SAUCE, two stochastic optimization frameworks for nonconvex learning under realistic noise assumptions, covering unbounded variance and Blum-Gladyshev noise. Established theoretical convergence guarantees and demonstrated their relevance to modern deep learning settings where classical bounded-variance assumptions fail.
• Currently investigating the optimization and generalization behavior of deep neural networks, with a focus on sharpness-aware training dynamics and adaptive scheduling.

Trustworthy Deep Learning, Aug 2024-Present

• Proposed a two-stage coreset selection algorithm with the dual purpose of eliminating spurious correlations and reducing dataset size. The deployed product achieved up to 27% improvement over standard ERM while utilizing only 10% of the training data on benchmark datasets such as Waterbirds, CMNIST, MetaShift, and UrbanCars, while also outperforming state-of-the-art invariant learning methods.
• Currently studying the training dynamics of deep learning models, with a focus on how loss landscape sharpness interacts with common training interventions and influences memorization and generalization, particularly in out-of-distribution settings.

Databoss Security & Analytics Inc., Data Scientist

Ankara, Turkey

Hourly Wind Energy Prediction, Jun 2023-May 2024

• Predicted the hourly energy data of multiple wind turbines in Turkey and Europe, showing chaotic and nonstationary behavior in multiple regions.
• Developed a large-scale ML framework consisting of various adaptive feature construction and selection methods with models such as deep learning models, machine learning models, statistical models, and state-of-the-art ensembling techniques. The deployed product increased short-term forecasting accuracy by an average of 21% and long-term forecasting accuracy by an average of 8%.

Natural Gas Demand Prediction, Aug 2021-May 2023

• Predicted daily international natural gas demand in Turkey, consisting of nonstationary multivariate time series from multiple sources.
• Developed a large-scale ML framework tailored toward a specific customer profile, where predictions are obtained based on key indicators automatically extracted from the data itself and/or user-given side information. The constructed framework was deployed to various natural gas production companies.

METU Center for Image Analysis (OGAM), Machine Learning Engineer

Ankara, Turkey

UAV Small Object Detection and Tracking, Jun 2020-Jul 2021

• Developed a Python application for real-time small target tracking by leveraging temporal data derived from object detector outputs, achieving 6% superior performance over the widely known YOLOv3 algorithm.
• Constructed a novel multi-hypothesis tracking algorithm with an annotation tool for semi-automatic correction of mislabeled UAV data. The deployed product decreased the total mislabeled data in various datasets by an average of 13%.

Academic Awards

Elmore Family School of Electrical and Computer Engineering, Purdue University

2024-2029

• Awarded the highly competitive and prestigious Elmore Fellowship Award for my graduate studies, given only to the most outstanding applicants.

Turk Telekom & Information and Communication Technologies Authority

2023-2024

• Awarded 5G and Beyond Joint Graduate Support Programme, a merit-based fellowship of monthly stipend during M.Sc.

Scientific and Technological Research Council of Turkey

2021-2023

• Awarded Directorate of Science Fellowships and Grant Programme, a merit-based monthly stipend and accommodation support during M.Sc.

• Received the 373rd rank among 2M high school graduates in the University Entrance Examination.

• Received the 11th rank among 0.2M university graduates in ALES, the National GRE.

Academic Duties

Reviewer Duties

Conferences: ICLR ‘25-‘26, NeurIPS ‘26, AISTATS ‘26, AAAI ‘25
Journals: IEEE TAI ‘25, IEEE IOT ‘24-‘25

Teaching Assistantship

Electrical and Electronics Engineering, Bilkent University, 2021-2024

• EEE321 Signals and Systems
• EEE202 Circuit Theory
• MATH255 Probability and Statistics

Software Skills

• Python: Professional research and industrial experience based on machine learning. High knowledge and experience in libraries such as PyTorch, TensorFlow, Pandas, NumPy, and Scikit-Learn. I have experience with Docker and Git via industrial projects.
• MATLAB: Professional research experience in signal processing and computer vision. Used for projects during courses, research, and internships.
• R: Intermediate-level experience, employed during research based on time-series forecasting using statistical models.
• C/C++: Intermediate experience in various course projects.

Languages

Turkish: Native
English: Conversationally Fluent [TOEFL IBT 108/120 (October 19, 2022)]