Paschalis Mpeis

A: Office 1.19/17, Informatics Forum, 10 Crichton Street , University of Edinburgh

E: paschalis.mpeis αt ed.ac.uk W: paschalis.mp

RESEARCH INTERESTS

The rapidly evolving architectures on computer systems has made the task of application optimisation non-trivial. There is no single set of optimisations that can yield optimally to any application on any architecture. Instead, a voluminous optimisation space has to be searched, restricting the fine-tuning only to particular pairs of application-architecture.

Iterative (or adaptive) optimisation is a technique that undertakes the task of exploring such optimisation spaces so as to improve various aspects of an application, i.e. performance, power consumption, code size, etc. It requires a high number of compilations and instrumented executions, which restricts its applicability to specific fields like embedded applications, i.e. optimise once, execute forever.

In particular, for mobile devices, that have restricted processing power and finite power sources, applying such a technique is prohibitively expensive. My research will focus on making this technique viable on such delicate environments.

EDUCATION

PhD in Informatics, University Of Edinburgh

Edinburgh, Scotland, UK • October 2014 – current

• My supervisors are Hugh Leather and Pavlos Petoumenos
• Member of the Institute for Computing Systems Architecture
• Member of the Compiler and Architecture Design Group

MSc in Computer Science, University Of Edinburgh

Edinburgh, Scotland, UK • September 2013 – August 2014

• Class of award: with Distinction
• Courses were focused on: High Performance Computing, Databases (Distributed, Big Data), and Distributed Systems
• For my Master’s thesis I have developed a crowdsourced iterative compilation system for Android devices. The project’s main challenges were the obtrusive iterative compilation technique and the finite sources of mobile devices.

BSc in Computer Science, University Of Cyprus

Nicosia, Cyprus • September 2009 – May 2013

• GPA: 7.58 out of 10
• Earned highest grade in Parallel Processing, Object Oriented Programming, and programming exercises at Data Structures and Algorithms
• Among the highest grades at Computer Architecture, Systems Programming, Internet Technologies, Operating Systems, Computer Graphics, Digital Image Processing and others
• Contributed to a project coined Vectormap (Spatio-Temporal Wi-Fi Localization), which was accepted for the final MOVE-COST conference, in Vienna. Additionally, a paper is currently under submission at a IEEE transaction journal.

High School

Paralimni, Cyprus • September 2001 – May 2007

• Distinction awarded each year from Paralimni Gymnasium and Lyceum

EMPLOYMENT AND EXPERIENCE

Intern at Data Management and Systems Laboratory • dmsl.cs.ucy.ac.cy

University of Cyprus, Nicosia, Cyprus • June 2012 – August 2013

• Smartlib Project • smartlib.cs.ucy.ac.cy

June 2012 – September 2012

Worked on FOSS project (Smartlib, a volunteer sharing library), which included a backend database and frontend web page and applications for mobile platforms: Android and iOS for volunteer book sharing, among faculty and students.

• Vectormap Project • tvm.cs.ucy.ac.cy

October 2012 – August 2013

Contributed to a project coined Vectormap, which provides fine-grained, low power and privacy-preserving indoor localization on smartphones. The backend consists of 2 parts: Hadoop HBase database on top of HDFS, and Oracle’s Glassfish server. The frontend is an Android application, which allows users to anonymously localize themselves indoors, while monitoring server’s speculations about their actual location.

Intern at Mobile Devices Laboratory • mdl.frederick.ac.cy

Frederick University, Nicosia, Cyprus • June 2013 – August 2013

• CYMAR Project • cymar.paschalis.mp

June 2013 – August 2013

Designed and developed a system for CYMAR Market-Research Ltd. By following a custom designed protocol, the company’s ITs can write and deploy dynamic questionnaires on Android tablet devices without the need of programming for Java/Android SDK. A handful of the supported functionalities are: mathematical operations, logical operations, nested branches, custom OO variables, and questionnaires flow/lifecycle control. The research staff use the dynamic questionnaires offline. On a subsequent stage the data are securely harvested on company’s servers. Progress reports of the research staff are automatically generated in order to be examined by the managers/ITs.

COMPUTING SKILLS

INTERESTS

I like underwater fishing, watching and playing football, and playing badminton. I also like listening to music, watching movies, and latest and greatest hacking my smartphones and tablets.

Last updated: 3 Oct. 2014