Jobs

Highlight Information

  • Students enrolled in a Master Degree in the academic year 2018/2019 can also apply for a position in this call
  • Duration of the PhD contracts: 4 years
  • Estimated starting date: March 2019 (possibly a few months later for candidates currently enrolled in a Master Degree)
  • If possibly interested, contact us ASAP by email at datalab@uca.es and we will help you navigate through the application process.
  • NEW EXTENDED DEADLINE: 14/01/2019

Description

We seek a qualified candidate to join our team working on Industry 4.0 projects with relevant companies like Airbus D&S. The contract has an initial duration of 6 months, but can be renewed up to three years.

The selected candidate will be part of the UCADatalab group and has the option to enroll on a Doctorate Program at Universidad de Cádiz, using part of his work towards completion of a PhD.

Qualification: BSc/MSc in Computer Science or equivalent studies.

Experience: Proven experience as a web developer

Required technologies:

– Django Framework (Python)

– Database management: MySQL or Postgre

– Front-end technologies: HTML5, CSS3, Javascript, Bootstrap/Foundation/SemanticUI or similar, JQuery, Json, AJAX

Desirable technologies:

– Additional back-end languages (Java, NodeJS, PHP)

– Additional front-end technologies (e.g. AngularJS)

– NoSQL databases (MongoDB, Neo4J, HBase)

– Development and deployment of RESTful architectures (DJango Rest Framework, Express, Jersey)

– Hadoop ecosystem

– UNIX system management

Salary: 1633 Eur/month (before tax). 2000 Eur yearly bonus by incentives.

 

Interested candidates are encouraged to send their CV to datalab@uca.es.

Description

We seek a qualified candidate to join our team working on Industry 4.0 projects with relevant companies like Airbus D&S. The contract has an initial duration of 6 months, but is expected to be renewed up to three years.

The selected candidate will be part of the UCADatalab group and has the option to enroll on a Doctorate Program at Universidad de Cádiz, using part of his work towards completion of a PhD.

Qualification: BSc/MSc in Mathematics, Physics or Engineering.

Required skills:

  • knowledge of python libraries for data science & machine learning (pandas, numpy, seaborn, scikit-learn)
  • data crunching and exploratory data analysis
  • supervised learning: classification and regression models.
  • visualization tools in python environments (matplotlib, seaborn, plotly, bokeh, etc.)

Desirable skills:

  • Bayesian models in statistics
  • Deep Learning frameworks (torch, keras, tensorflow)
  • Time series forecasting
  • Big data ecosystems

Salary: 1633 Eur/month (before tax). 2000 Eur bonus by incentives.

Interested candidates are encouraged to send their CV to datalab@uca.es.

PhD positions

Highlight Information

  • Students enrolled in a Master Degree in the academic year 2018/2019 can also apply for a position in this call
  • Duration of the PhD contracts: 4 years
  • Estimated starting date: March 2019 (possibly a few months later for candidates currently enrolled in a Master Degree)
  • If possibly interested, contact us ASAP by email at datalab@uca.es and we will help you navigate through the application process.
  • NEW EXTENDED DEADLINE: 14/01/2019

Dimensional control of large structures using artificial vision techniques

Description

Large ships are built modularly by joining large blocks that are built separately. Due to its size, the existence of dimensional deviations is inherent to their manufacture. If dimensions are not controlled or are innacurate, failures in the subsequent assembly of the parts can have a large cost effect and temporal delays that can affect severely the shipyard’s productivity. Moreover, most of the techniques available nowadays are not flexible and the result of the measurement is usually a discrete number of data that may not be sufficiently representative of the details of large three-dimensional pieces. In this thesis will be apply and develop flexible metrological techniques that would provide accurate models in three dimensions (3D) of the large structures that are typically found in a shipyard. With this aim, artificial vision techniques will be used to facilitate the capture of the spatial coordinates of a large number of keypoints, all that in a semi or automatic way. Different modes and devices for image data acquisition (cameras, mobile, drones, lasers…) will be evaluated. The data will be processed using a combination of image processing algorithms in 2D and 3D.  And the 3D models will be tested against CAD models to be able to evaluate with immediacy deviations with respect to the design.

Candidate Profile

Engineer or Master Degree in Engineering (Computing, Robotics & Automation, Physics)

Resilient security solutions in ad hoc networks: a survivable approach

Description

Nowadays, most of people are living in a dynamic world where things and humans are unavoidable interconnected. Some real examples of such a kind of fully connected worlds are the smart cities or the smart grids which are in turn supported by underlying technological architectures and devices. In this types of scenarios, the ad hoc networks suit well due to the special characteristics being, however, the last ones the main cause of their security weaknesses. For example, the inherent decentralized architecture and self-management characteristics make them vulnerable to identity spoofing (sybil), data manipulation or fabrication (data tampering) or dropping attacks, among others. Depending on the context of application, the previous attacks could affect not only on the services offered by the system but on humans lives.

In this context, security solutions based on the use of prevention, detection or response/tolerance mechanisms have been proposed, being the last ones less frequent. It is a fact that most of them are focused on fighting against certain types of attacks and usually acts on specific parameters and/or characteristics of the involved system. This approach, on one hand, makes the proposals more robust and effective for the threats they were designed to but, on the other hand, they are definitively less resilient/effective when new security threats or attacks appear. In the context of the security in information and communication technology (ICT) such kind of attacks are known as zero-day attacks.

All the previous reasons make necessary the proposal of novel security solutions and mechanisms addressing large amount of requirements, characteristics and parameters from a global point of view. All together, will provide more robust and resilient systems, in other words, survivable systems against unknown security threats or attacks.

Certainly, it is a great and open research challenge for the security on ICT systems that would add the necessary capabilities in the mitigation of the impact of almost all attacks on the system where they were deployed.

Candidate Profile

  • Telecommunications or Computer Science engineering/degree.
  • Cybersecurity background. Msc. in Cybersecurity is recommended
  • Good knowledge in communications networks and systems.
  • High motivation in working on the ICT security field.
  • Good reading and writing in English.
  • Good academic background.

Internet de las Cosas para la Industria 4.0

Description

Esta tesis doctoral pretende explorar la aplicación de IIoT (Industrial IoT) en el marco de la denominada cuarta revolución industrial (Industria 4.0), proponiendo nuevas técnicas y tecnologías para resolver problemas industriales en colaboración con Airbus D&S. Durante la duración de la tesis, el candidato/candidata adquirirá y desarrollará conocimientos en:
• Plataformas IoT de aplicación, eminentemente industrial, para la interconexión de dispositivos, recogida y gestión de información
• Uso y despliegue de infraestructuras IoT de propósito general como LoRa o NB-IoT
• Puesta en marcha de ecosistemas IoT aplicables en el ámbito industrial
• Sistemas de autenticación y autorización para cumplir criterios de seguridad en el ámbito industrial
El candidato abordará el uso de tecnologías y técnicas relacionadas con el objetivo de encontrar soluciones en los retos presentes en IIoT: Conectividad, Diseminación, Almacenamiento de datos, Seguridad e Integración. Para esto, y como resultado de estas tesis, se pretenden encontrar nuevas soluciones que, además de generar un conocimiento relevante, sean de aplicación y utilidad directa para Airbus D&S.

Al finalizar esta tesis, el/la candidato/candidata adquirirá tanto una formación académica y tecnológica así como una experiencia profesional en uno de los más demandados campos de aplicación actual, como es la Industria 4.0

Data Science and Machine Learning in Industrial Applications

Description

The PhD thesis will explore the application of data science and Machine Learning algorithms to solve real world industrial problems in collaboration with Airbus D&S. During his postgraduate training, the candidate will acquire and develop knowledge of:
  • simple methods for supervised learning: Random Forest, Factorization Machines, SVMs, etc.
  • deep learning (LSTMs, Conv-Nets)
  • forecasting in time series (DLMs, Kalman filters).
  • Bayesian methods
  • techniques for data cleaning, missing data imputation and visualization.

The candidate will tackle applied ML projects on computer vision (object detection, image classification), time series forecasting or Natural Language Processing, with special emphasis on their application to industrial problems.

By the end of the thesis, the candidate will have acquired both a valuable technological formation and professional experience in one of the most demanded profiles in the market.

Candidate Profile

  • Masters degree (currently doing or completed) in Mathematics, Physics, Computer Science or Engineering.
  • Good academic background.
  • Competent coding skills.
  • Advanced knowledge of english.
  • Entrepreneurial character and interest for applied problems.

Design of intelligent systems based on domain-knowledge information

Description

In the last years, artificial intelligence and machine learning models have been used as a powerful tool to develop intelligent systems which allow to extract and learn patterns from data in order to obtain accurate predictions in new future samples. Machine learning models usually perform better when enough samples are available to train these models. In this sense, the volume, variety and speed for which data is nowadays generated in the era of Big Data helps to build highly accurate predictors. However, there are still studies in different areas in which the volume of data available is either low or the process of data collection is difficult and/or expensive. Therefore, there is a need of modifying the models in such a way that domain-knowledge information is somehow used to impose restrictions to the training process with the aim of helping to overcome overfitting issues and achieving higher generalization rates. To date, some works are already available that propose two linear models (Group Lasso and Sparse Group Lasso) using domain-knowledge information to group the input features into different groups [1-2]. On the other hand, in [3] domain-knowledge information is used within the fitness function of a genetic algorithm in order to guide the search of the most relevant variables that maximizes the accuracy achieved by a set of models considered in that study. More recently, in [4] problem-specific information published in hundreds of related previous studies is integrated to control the degree of regularization applied to the input features of a linear model outperforming a baseline linear model with homogeneous priors. Despite the advances introduced by these works, there is a limitation of using only linear models which are unable to learn nonlinear relationships between the input features and the event of interest. Hence, the main research line of this PhD thesis focuses on the incorporation of domain-knowledge information in more sophisticated machine learning models as well as on the application of novel synthetic data generation techniques in order to minimize the risks of overfitting issues which may exist in research areas with a small data volume.

 

[1] Meier, L., van de Geer, S., Bühlmann, P.: The group LASSO for logistic regression. Journal of the Royal Statistical Society Series B (Statistical Methodology) 70(1):53-71 (2008)
[2] Simon, N., Friedman, J., Hastie, T., Tibshirani, R.: A sparse-group lasso. Journal of Computational and Graphical Statistics 22(2), 231–245 (2013)
[3] Luque-Baena, R.M., Urda, D., Claros, M.G., Franco, L., Jerez, J.M.: Robust gene signatures from microarray data using genetic algorithms enriched with biological pathway keywords. Journal of Biomedical Informatics 49 (2014), doi: 10.1016/j.jbi.2014.01.006
[4] Urda, D. et al: BLASSO: Integration of biological knowledge into a regularized linear model. BMC Systems Biology, 12(S5) (2018), doi:10.1186/s12918-018-0612-8

Candidate Profile

– BSc, MSc in Computer Science; BSc, MSc in Telecommunications Engineering; BSc, MSc in Industrial Engineering; BSc, MSc in Mathematics; or similar degrees.
– Good programming skills.
– Good level of English (reading/writing).