Evangelos Stromatias, Phd Email & Phone Number

Berlin, Germany

Collaboration with data scientists, data engineers to design state-of-the-art infrastructure to serve machine learning models in production. Furthermore, I developed, productised and maintained ML models using FastAPI and Celery.Infrastructure:• Introduced and established best practices with regards to programming, code and system documentation, Architectural Decision Records (ADRs) and CI/CD pipeline standards (Azure). • Implemented alerting for our ML services by introducing… Show more Collaboration with data scientists, data engineers to design state-of-the-art infrastructure to serve machine learning models in production. Furthermore, I developed, productised and maintained ML models using FastAPI and Celery.Infrastructure:• Introduced and established best practices with regards to programming, code and system documentation, Architectural Decision Records (ADRs) and CI/CD pipeline standards (Azure). • Implemented alerting for our ML services by introducing Kubernetes health probes and web-hooks.• Developed internal libraries and templates for the development of pipelines and APIs.Machine Learning contributions to the Product:Product Quality: • Developed, productised and maintained ML algorithms to assess the quality of submitted manuscripts the result of which was the increase in the quality of Frontiers reputation.Recommender Evaluation Framework: • Researched, developed and productised a framework to enable the comparison of Recommender Algorithms for internal and external use.Soft skills: • Organised and facilitated meetings with stakeholders to gather requirements for ML products. • Frequent meetings with stakeholders to assess the quality of an ML product and gather feedback. Show less

Leiden, Netherlands

Research and implementation of clinical software products in the field of Deep Learning (DL). More specifically:• Developed a DL pipeline in Python, on top of Keras & Tensorflow, with the purpose of having reproducible experiments, improving team collaboration, and reducing the time-to-market of segmentation/classification based products. • Experimented with Unet and Fully-Convolutional Networks for semantic segmentation in Keras.• Created material and facilitated DL Workshops… Show more Research and implementation of clinical software products in the field of Deep Learning (DL). More specifically:• Developed a DL pipeline in Python, on top of Keras & Tensorflow, with the purpose of having reproducible experiments, improving team collaboration, and reducing the time-to-market of segmentation/classification based products. • Experimented with Unet and Fully-Convolutional Networks for semantic segmentation in Keras.• Created material and facilitated DL Workshops for the applied research team. Show less

Amsterdam Area, Netherlands

Research and development for Computer Vision and Machine Learning tasks in order to bring innovative solutions to TV operators and Broadcasters. More specifically, I was involved in the following projects: • Empirically evaluated various Deep Face Descriptors for our face recognition pipeline• Researched and developed an automatic thumbnail suggestion system (Episodic Images) that suggests thumbnails to users given an input video which won an industry UX award in 2021.• Investigated… Show more Research and development for Computer Vision and Machine Learning tasks in order to bring innovative solutions to TV operators and Broadcasters. More specifically, I was involved in the following projects: • Empirically evaluated various Deep Face Descriptors for our face recognition pipeline• Researched and developed an automatic thumbnail suggestion system (Episodic Images) that suggests thumbnails to users given an input video which won an industry UX award in 2021.• Investigated and deployed visual attention models to automatically detect and propose “interesting” regions in images and videos with the purpose of detecting beginning/endings of TV program times and commercial breaks• Fine-tuned our logo recognizer model (Faster RCNN) to meet client needs• Supervised AI MSc students with their final year projects • Gave presentations at Meetups, facilitated brainstorming sessions, and problem-solving sessions within a Scrum team• Developed microservices in Java and Python using Prometheus for monitoring• Carried out basic sysadmin tasks such as: setting up and updating computer nodes with Puppet and interface code with Cassandra databasesWorked mostly with Python and the following libraries: OpenCV, DLIB, Numpy, Scipy, Scikit-Learn, Keras, Tensorflow, Caffe, Pandas, matplotlib, Flask Show less

Delft Area, Netherlands

Visiting researcher in the Pattern Recognition and Bioinformatics group as part of the ongoing collaboration with Samsung’s Neural Processor Project

Sevilla Area, Spain

I was part of the Neural Processing Project; a global project funded by Samsung with the purpose of investigating how to accelerate efficiently Deep Neural Networks on mobile platforms with limited resources. More specifically:• Worked with event-driven Convolutional Neural Networks and devised a novel method to train a supervised classifier using Theano• Investigated novel unsupervised methods to extract features from asynchronous inputs and validated their performance with various… Show more I was part of the Neural Processing Project; a global project funded by Samsung with the purpose of investigating how to accelerate efficiently Deep Neural Networks on mobile platforms with limited resources. More specifically:• Worked with event-driven Convolutional Neural Networks and devised a novel method to train a supervised classifier using Theano• Investigated novel unsupervised methods to extract features from asynchronous inputs and validated their performance with various Machine Learning algorithms using the Scikit-Learn library and Python (Patent Pending)• Used pandas to collect and analyze experimental results from multiple simulations• Developed an event-driven neural network simulator in C and Python Show less

United Kingdom

Our research group was part of the Human Brain Project, which included many research institutes and universities across Europe. My Ph.D. research took place on a novel many-core ARM-based embedded platform where I:• Trained, developed and deployed event-driven Deep Neural Networks for Computer Vision tasks• Investigated the performance and robustness of event-driven Deep Neural Networks in terms of power requirements, execution speed and bit resolution for model parameters•… Show more Our research group was part of the Human Brain Project, which included many research institutes and universities across Europe. My Ph.D. research took place on a novel many-core ARM-based embedded platform where I:• Trained, developed and deployed event-driven Deep Neural Networks for Computer Vision tasks• Investigated the performance and robustness of event-driven Deep Neural Networks in terms of power requirements, execution speed and bit resolution for model parameters• Interfaced novel Neuromorphic sensors such as Silicon Retinas and Cochleas• Developed software in Python and embedded C to enable users to describe, deploy and simulate Neural Networks in our platform. This code base was shared between collaborators around the world• Organized and led groups in several scientific workshops, demonstrated ability to work effectively in interdisciplinary team environments, and gave numerous oral and poster presentations at international conferences Show less

United Kingdom

During my Ph.D. I was a teaching assistant in the following courses: • Mobile Systems (COMP61232) • Optimization for learning, planning and problem-solving (COMP60342)• Fundamentals of Computer Engineering (COMP12111).

Aspropyrgos Refinery, Greece

Received training in:The refinery’s control systems, with emphasis given to Yokogawa’s DCS – Centum V, CS,and CS 300 where we dealt with hardware and software issues.The Emergency Shutdown (E.S.D) system of the refinery, which included several P.L.C.brands, such as Siemens S5, S7, Allen Bradley and Hima. Control loops and sequence tables.Control valves and transmitters (as of level and pressure). Wiring and instrumentation.

Doctor Of Philosophy (Ph.D.), Computer Science, Aii

Investigated the scalability and performance of spiking neural networks on SpiNNaker, an ARM based massively parallel architecture.

Master’S Degree, Microelectronic Systems, Distinction

Activities and Societies: Liverpool Band Society Dissertation title: “Developing a supervised training algorithm for limited precision.

Bachelor’S Degree, Automation Engineer

Dissertation title: “Development of a photovoltaic model and maximum power point tracker using Matlab/Simulink”, 2010. The project.