I am currently working in the AI Lab at British Antarctic Survey participating in Marine operation planning in the polar regions. A project that aims at leveraging ML to efficiently operate the research vessels and resources in complex environment conditions. I am specifically involved in the efficient Ocean modeling/digitization and green route planning modules of the project.

I hold a PhD from the University of Cambridge, Computer Science department. My research experience revolves around machine learning and systems. I worked in my PhD on optimizing big data systems and accelerating big data analytics using Machine learning. This work has led to “Tuneful”, a system that leverages incremental sensitivity analysis and Bayesian optimization to efficiently tune the configurations of big data systems, saving the search time by 2.7X and implemented as an extension to Apache Spark.

I have also proposed SimTune, an efficient similarity-aware configuration tuner for big data analytics. It leverages neural encoding of workload execution metrics to detect workload similarity, then shares the tuning knowledge between similar workloads using Multitask Bayesian optimization. SimTune reduces the search time for close-to-optimal configurations by 56-73% when compared to existing state-of-the-art techniques.

More details about my PhD work can be found in my KDD’20 paper “To Tune or Not To Tune? In search of optimal configurations for data analytics” (acceptance rate <6%). During the course of my PhD, I have successfully secured $63K of funding awards (gifted as cloud research credit) to experiment with deploying my work on Google and Amazon cloud infrastructure.

Before starting my PhD, I worked at Orange labs as a Software Engineer for 3 years and an R&D Engineer for 2 years, and at Intel as a system validation engineer for one year, where I gained a broad range of industrial experience and participated in prototyping a wide array of Mobile and M2M development and research projects. I also worked as a teaching assistant at Cairo University, where I taught several computer science courses and did my masters in the optimization of energy consumption in mobile computing systems using machine learning (SVM). This work led to “Smartphone Energizer”, which is explained in detail in my master’s paper here.

Research interests

• Data-Efficient learning
• Applied ML
• Optimization of data intensive scalable systems (DISC)
• Meta-learning
• Energy-aware Computing

• Ayat Fekry, Lucian Carata, Thomas Pasquier, Andrew Rice, and Andy Hopper. “To tune or not to tune? in search of optimal configurations for data analytics“. KDD’20.
• Ayat Fekry, Lucian Carata, Thomas Pasquier, Andrew Rice, and Andy Hopper. “Accelerating the Configuration Tuning of Big Data Analytics with Similarity-aware Multitask Bayesian Optimization”. IEEE BigData’20.
• Ayat Fekry, Lucian Carata, Thomas Pasquier, Andrew Rice, and Andy Hopper. ”Tuneful: An Online Significance-Aware Configuration Tuner for Big Data Analytics.” arXiv preprint. arXiv:2001.08002 (2020).
• Ayat Fekry, Lucian Carata, Thomas Pasquier, Andrew Rice, and Andy Hopper. “Towards Seamless Configuration Tuning of Big Data Analytics“. ICDCS’19.
• Fekry, Ayat. “Big Data Gets Bigger: What about Data Cleaning Analytics as a Storage Service?.”. 9th USENIX Workshop on Hot Topics in Storage and File Systems (HotStorage 17). USENIX Association, 2017.
• Khairy, Ayat, Hany H. Ammar, and Reem Bahgat. “Smartphone Energizer: Extending Smartphone’s battery life with smart offloading.”. Wireless Communications and Mobile Computing Conference (IWCMC), 2013 9th International, pp. 329-336. IEEE, 2013.

Publications from NERC Open Research Archive

2. Logist AI for environmentally aware decision support

   AMOP is developing Artificial Intelligence methods that aim to optimise the efficiency of Antarctic field operations, while maximising science delivery

   Read more of: Logist AI for environmentally aware decision support

   

• Digital Twin of the Polar Regions
• AI for carbon reduction in polar navigation