Contact: (646) 496-6693 | hw2738@columbia.edu | LinkedIn
Online Buffer Management with Predictions [1]
Columbia University
New York, NY
Sep 2022 – Apr 2023
• Explored online buffer management, incorporating predictive future arrival.
• Attain a 7.2% improvement over the current best algorithms on both synthetic and real datasets.
• Developed pioneering theoretical frameworks with simple greedy strategies, utilizing Python libraries (NumPy, pandas) for experimental analysis.
Energy-Efficient Scheduling with Predictions (Accepted by NeurIPS ‘23)
Columbia University
New York, NY
Jul 2022 – Feb 2023
• Devised an algorithm for enhancing energy efficiency in online scheduling, leveraging machine learning predictions.
• Achieved a 9.1% improvement over the current best approach on a real dataset.
• Utilized advanced combinatorics techniques, employing tools such as Python's NumPy libraries, to analyze SNAP datasets.
Online TSP with Predictions [2] (Under review by INFORMS J. Comput.)
Columbia University
New York, NY
Feb 2022 – Jul 2022
• Enhanced efficiency of the online Travel Salesman Problem by incorporating future demand predictions.
• Obtained a 10% empirical improvement over the current best algorithm on a real dataset.
• Implemented algorithms based on greedy strategies, utilizing Python's NumPy libraries for data analysis.
Scheduling with Speed Predictions [3] (Accepted by WAOA ‘23)
Columbia University
New York, NY
May 2021 – Feb 2022
• Investigated a scheduling problem integrating uncertain machine speed predictions.
• Achieved an average 16.7% improvement over the current best result on synthetic datasets.
• Applied recursive and divide-and-conquer techniques in algorithm design, leveraging Python for experimental analysis.