Table of Contents
- Our speakers
- Tech Track sessions
- Agile and Business Track Sessions
- In the spotlight Ghelamco alert
- In the spotlight Some pitfalls of AI
As is annual tradition, even despite Corona, we held our 8th JOIN event in the beginning of October. JOIN stands for JWorks Open Innovation & Networking and aims to bring interesting talks on a wide variety of topics. This years edition was a bit different being that large gatherings of people are a no-no, so we opted for an online only edition.
We stuck to our dual track approach to provide some interesting sessions for everybody:
- A Tech track consisting of 4 sessions:
- Ghelamco alert: Combining Raspberry Pi and AWS Cloud to get the best of both worlds - Bas Moorkens
- Infrastructure is code with the AWS Cloud Development Kit (AWS CDK) - Kevin Azijn
- Moving a complex and slow deployment pipeline to a streamlined and lightning fast Azure deployment - Pieter Vincken
- Some pitfalls of AI - Joachim Ganseman
- An agile and business track also consisting of 4 sessions:
- Dancing the BOSSA nova – how to bring a culture of experimentation into your company - Edwin Burgers & Maryse Meinen
- The power of Kata - Michaëla Broeckx
- Failure Culture – what it means to fail and how we can gain value from it - Anke Maerz
- Why don’t you get my code? - Francis Laleman
Read and watch on down below as we go over each speaker and talk.
Like any event we depend on our contributors, in our case being our speakers. Without them our event would not be possible, they provide us with the awesome and interesting sessions.
Bas is a cloud platform architect at Ordina Belgium who is fascinated by AWS, containers and pipeline automation. He started off as a Java full stack developer but got more and more into the DevOps and cloud world where he currently spends most of his time.
Company: JWorks - Ordina Belgium
After spending more than 10 years as a Software Engineer and Tech Lead in both consulting and financial world, Kevin got introduced to public cloud and started building web/mobile solutions and managing DevOps teams on AWS for the Flemish Government. After several years the call of technology over management became too loud to ignore and Kevin decided to join AWS as a Solutions Architect. At AWS Kevin is supporting customers with building a vision and architecting for the future.
Company: Amazon Web Services
Pieter Vincken is a Cloud Automation Engineer with a strong interest in anything related to Cloud Native. He likes to optimize development workflows, from Ideation until code running in production, by enabling CI/CD to be fully automated. Any solutions he creates, will have started as an architectural drawing.
Company: JWorks - Ordina Belgium
Joachim Ganseman is a computer scientist and has a history as a PhD student at the University of Antwerp, with side jumps to Queen Mary University in London and Stanford University, focusing on digital signal processing, machine learning and audio analysis. Since 2018, he has been working at Smals Research where he focuses on AI-related topics, including Natural Language Processing and Conversational Interfaces, and their potential applications in governmental contexts. In addition to his work, he is an excellent pianist, and as co-founder and organiser of the Belgian Informatics Olympiad, he received the annual prize for science communication from the Royal Flemish Academy of Sciences of Belgium in 2016.
Company: Smals Research
Edwin Burgers & Maryse Meinen
Edwin is a Scrum Master, agile coach and leader with 10+ years in the agile field. He has supported many teams and organisations to improve value delivery. To survive in the current highly competitive world agile frameworks, practices and leadership are not enough. Less hierarchy, more autonomy and a higher involvement from the people on the floor in strategy and decision making are crucial. That´s why he is focussing on ways to address these aspects in his work as a coach or leader.
Maryse is a 100% scrum master, helping people and organisations increase their agility, so that they can reduce their (organisational) suffering. She coaches teams and executives in solving their own challenges and impediments in their work. As a true corporate rebel, the Socratic method for better conversations is key in her approach. More ancient philosophy should be compulsory in the agile world. And she is not dogmatic: all should use scrum.
Maryse and Edwin worked together at Ordina from 2015 and started their own business in Practical Agile in 2018.
Company: Practical Agile
Michaëla is an Agile coach with a focus on value and the human aspects of work. She is devoted to Agile & Scrum, not just as a mindset and framework for collaborative product delivery, but also for their general quality as ways to unlearn innate/inbred habits that prevent us from learning efficiently. A true Agile mindset enables enterprises to break down siloes, and build professional human networks throughout the company and beyond.
Company: AgileWorks - Ordina Belgium
I am Anke and I have a mission in my life: To see and enjoy beauty in life, and to help other people to see and enjoy it, too.
As a Scrum Master and Agile Coach I live this mission by asking myself every morning THE question: What is it that keeps these amazing people from achieving their goals? – And whatever my answer is, that is where I want to support. Because that is where I can create space for them to see and enjoy the beauty of life.
This has proven very purposeful to me since 2014, when I first got introduced to Agile – and through various big and huge companies in Belgium as well as Germany.
Before getting into Agile I’ve been project manager and requirements engineer for 2 years, and before that I studied mathematics and psychology at the university of Tübingen (crazy mix? YES! Beautiful, right?!!)
Even before that time I’ve already been practicing to be a human, which has proven to be most useful on a personal as well as professional level!
Company: AgileWorks - Ordina Belgium
Today, I mainly work as a designer of cooperative learning processes and learning cultures. I am a gentle facilitator of (Agile) transformation processes, a train-the-trainer, a cooperative learning facilitator and an Agile coach. Having spent so much time in the Middle East and in the Indian subcontinent, I have now been turning my attention to Japan - studying TPS, Lean and Kata, but also the less obvious, such as ikigai, Taoism and Zen Buddhism, or the shuhari learning process in traditional Japanese craftsmanship. Most probably, however, you will find me in the garden.
Company: Agile & Learning Beyond Borders
I would like to thank our speakers once more from the whole of Ordina for the interesting sessions they have given.
Tech Track sessions
Session 1 - Ghelamco alert: Combining Raspberry Pi and AWS Cloud to get the best of both worlds - Bas Moorkens
In this talk we will present our Ghelamco alerting solution which runs on a Raspberry Pi and is empowered by the AWS cloud to provide additional capabilities. Come discover how we turned our RPI into a managed device by leveraging the AWS cloud. We will demonstrate the power of this technology set by first covering the scenario where the solution is not connected to the internet. In the second part we will do a deep dive on the scenario where the RPI is connected to the internet. In this part we will show you how we can leverage more services of the AWS cloud to bring out the full potential of this solution, effectively giving us the best of both worlds.
Session 2 - Infrastructure is code with the AWS Cloud Development Kit (AWS CDK) - Kevin Azijn
The AWS Cloud Development Kit (AWS CDK) is an open source software development framework to model and provision your cloud application resources using familiar programming languages. Provisioning cloud applications can be a challenging process that requires you to perform manual actions, write custom scripts, maintain templates, or learn domain-specific languages. AWS CDK uses the familiarity and expressive power of programming languages for modeling your applications. It provides you with high-level components that pre-configure cloud resources with proven defaults, so you can build cloud applications without needing to be an expert. AWS CDK provisions your resources in a safe, repeatable manner through AWS CloudFormation. It also enables you to compose and share your own custom components that incorporate your organization’s requirements, helping you start new projects faster.
Session 3 - Moving a complex and slow deployment pipeline to a streamlined and lightning fast Azure deployment - Pieter Vincken
How to automate application platform deployments. Tips and tricks about a migration from an on-premise solution onto Microsoft Azure. Want to know how we went from 6 week lead times on a development environment to only 90 minutes and a single push of a button?
Session 4 - Some pitfalls of AI - Joachim Ganseman
The hype surrounding AI and machine learning creates high expectations. There is a strong focus on software that ultimately has to make autonomous decisions and relies on difficult to access and complex statistical conclusions from large amounts of data. This comes with some risk: errors can creep in all steps from design to rollout, which can thoroughly mess up the end result. In addition to the spectacular AI success stories of recent years, there were also quite a few spectacular mistakes. In this webinar we highlight that downside: what can go wrong in an AI project and what should you pay attention to? The agenda includes: bias and fairness, adversarial attacks, explainability, as well as concerns about security, privacy and ethics.
Agile and Business Track sessions
Session 1 - Dancing the BOSSA nova – how to bring a culture of experimentation into your company - Edwin Burgers & Maryse Meinen
In the VUCA world that we are in, companies are expected to be flexible and both rapidly responsive and resilient to change, which basically asks them to be agile. The authors of the book BOSSAnova, Jutta Eckstein and John Buck saw 4 recurring problems at the agile transformations they were involved in, where they saw that all that comes with the agile frameworks, principles and practices is not sufficient for company-wide agility. They combine Beyond Budgeting, Open Space, Sociocracy and Agile in practical guide for creating company-wide agility, in a probe-sense-respond manner. This workshop gives a short introduction to BOSSAnova and how it can help to start probes and experiments in your own organization. With an approach and a lot of good practice experiments done in other organizations.
Session 2 - The power of Kata - Michaëla Broeckx
We will jointly look for efficient ways to convert good intentions into value, by investigating some personal challenges and how we succeeded or failed. We will examine these techniques and connect them to how our brain works. For every challenge we take on, our human tendency to take the easiest path (the one we imagine to be right) can be counterbalanced by deliberate scientific thinking.
Session 3 - Failure Culture – what it means to fail and how we can gain value from it - Anke Maerz
I don’t like failing. I want to achieve a goal, I invest my time and energy, I am looking forward to achieving this goal, I am getting closer, closer - and fail. The goal is not achieved and I want to retreat into a dark corner and just forget. This is a perfectly normal human reaction. Yet, as we all know, no big endeavor happens without its fair share of failure - some huge achievements even happen out of failure alone… Hence, there is something useful to it. Something that even smells like success! This talk analyses different projects, their main impediments and success or failure factors, includes some interactive parts - and in the end (hopefully) leaves us with some renewed courage to face the failures in our lives… to gain value from them.
Session 4 - Why don’t you get my code? - Francis Laleman
Video under review - will be added soon!
This is a playful, interactive session on the concept of “code” - as a formal, grammar-and-syntax-driven substitute for “meaning”. We all know that code is the central edifice of what it is that we are doing - at work, and, perhaps, even in life. But do we fully understand the profound relationship between code and meaning? Do we really grasp how code, in all its formality and conventions, both pins us down into the confinement of “knowns”, and has the power to liberate us from what is already known and drive us forward? After this session, expect to be aware of more than just “code”, and to be able to understand “meaning”, in life and work, through a variety of grammars and syntaxes, more diverse than ever imagined.
In the spotlight Ghelamco alert
The awesome internship project by Kevin Govaerts is a shining example of an end-to-end integrated IoT project. It touches a lot of interesting technologies and integrates everything very nicely and includes managing a Raspberry Pi device via AWS.
The project provides a solution to our parking problem at our Ghelamco office in Ghent. Since the parking lot needs to be vacated whenever there is a “home” soccer match or we get a hefty fine per car that remains, we needed an automated solution to warn the people at the office. This project does just that, end-to-end. It reads & scrapes the data and determines when the matches are and commands the Raspberry Pi to light up the alarm light.
It uses at least 10 different AWS services and the usage of AWS Greengrass allows you to deploy production ready applications to edge devices without having to manage it yourself. I wont go into all the technical details since you should absolutely read Bas and Kevin’s their excellent blog post about the project! Be sure to check out the blog post and the video of the project!
In the spotlight Some pitfalls of AI
As I myself am very interested in Machine Learning and Artificial Intelligence, this talk in particular peaked my interest. Machine Learning and AI are being portrayed as the silver bullet, as the thing every project needs. However developing a Machine Learning model/system is not as straightforward as one might think.
This talk goes into detail about what can go “wrong” and what the common pitfalls are when developing such a system. Datasets can be a tricky matter to fully grasp and get right. Hidden correlations or biases make that our trained models can often show the same discriminatory tendencies like normal people do. Removing biases from a dataset is hard, very hard. Even if you think you are not biased, please go on and go over the list of cognitive biases.
Removing certain data fields/variables from the dataset in order to prevent biases can have little to no effect due to the fact that there are hidden variables. The talk explains this with a great example, when a model has a gender bias one might think that removing the gender from the dataset might solve the issue. This might not work since other variables might indirectly propagate the same gender bias. If the model handles text, the gender is actually also encoded in the vocabulary and way of speaking. It is very different between women and men.
Machine Learning models are vulnerable to attacks:
- Data poisoning attacks
- When people can enter data or find a way to add data to a source that is being used to train a model.
- Verifying and cleaning data is very important to prevent this!
- Adversarial attacks
- Changing a few pixels in an image might throw off the model completely.
- Because the training dataset is so large the model becomes sensitive to small and very particular changes in the input.
These attacks makes using these models for very critical things like autonomous driving tricky and potentially dangerous. While this problem is not fully solved yet, the field is advancing at an extremely high tempo, so these type of problems pose less and less of an issue.
Machine Learning can be, and is used, for less legal & nice practices:
- Spear & laser phishing
- Phishing attacks that use very specific data collected on the target to believe extremely believable phishing vectors.
- Fake news becomes harder and harder to detect because images and even videos can be altered. Faces can be swapped, even audio can be changed to match the voice of someone else.
- This model can write full texts, code, layouts. Often the text that is generated makes more sense than anything a certain American president would say.
- Recommendation engines
- Watch a certain type of video, get suggested more of the same but more extreme. This enables a feedback loop, the type of suggested content is in your area of interest just a tad more extreme/strange/unbelievable, so you keep watching, and the next suggestions rinse and repeat the cycle. This is a big issue since it has a tendency to propagate extremes. What you see is not what other people see, it is curated for you by your past consumed content. More than ever it is extremely important to check facts and do some digging on independent sites/new sources.