During the presentation, we will predict if NASA has found exoplanets (planets outside the solar system). And, for this, we are going to use the results of the Kepler satellite and also IntegratedML, the Machine Learning module for InterSystems IRIS.

Distributed systems are inherently complex and exposed to failures in their components. While distributed algorithms try to do their best in handling node and network failures, it is often not possible to mask them. In the presence of failures, systems that thrive for high availability need to allow users to always submit operations and observe the local state (ensuring local-first), leading to potential divergence. Designs with Conflict-free Replicated Data Types (CRDTs) have been used extensively to guarantee that it is possible to converge to a state that reflects concurrent operations. Recent proposals show that it is possible to go beyond state convergence and use CRDTs as a basis for providing additional system-wide guarantees on individual data items, non-negative inc/dec counters, and cross-item guarantees such as referential integrity. This talk introduces the main concepts underpinning CRDTs and these recent proposals. 
 

“Cloud native?” Check! Apply the same principles at the Edge? Hmmm…! How do I operate Apps across 1000s of locations, which are often hidden behind layers of NAT? How do I run AI apps on nodes that are too small to fit the AI model? How to make it operationally simple? Let’s discuss and demo!

We’re all familiar with “cloud native” - but once we start to operate applications at the edge, we have to adopt a new set of principles and evolve our cloud-native paradigms. We deploy Apps at the edge to achieve lower latency or higher performance, to comply with data sovereignty regulations, to reduce transit cost or to perform near real-time decision making on local data sources.

Developing and operating Edge apps requires us to answer questions like:

• How do I operate Apps across 1000s of locations, which are often hidden behind layers of NAT and have spotty cloud connectivity?

• How do I run computation heavy tasks, like AI apps, on a set of nodes where each node does not have sufficient CPU and memory to run the entire model?

• How do I deal with a heterogeneous environment, with x86 and ARM-based devices?

• Which additional tools do I need to assure compliance to data-privacy rules, run AI models that just don’t fit a single compute element, or perform federated learning in an efficient way?

This session will address those questions and introduce the “edge native” paradigm. It will provide an understanding of how to design and operate Edge apps. A set of demos and example use-cases complements the conceptual discussion - making “edge native” a reality for the audience.

Dealing with failures is arguably the most important aspect of any system. Oftentimes, it is what stands between a product that runs as expected and one that keeps producing surprises and causing investigations. When done right, handling of failures is what differentiates a professional from an amateur.

In this talk Sergey Bykov will discuss pros and cons of the commonly used patterns of dealing with failures in distributed systems and cloud services, and the tradeoffs we inevitably make when designing such systems.

This talk will share our experiences migrating more than 300 ingestion processes from a traditional ETL approach to a brand new serverless architecture in AWS: the challenges and lessons learned along the process, how the new platform allowed us to boost developer efficiency and time to market while improving the performance of processes and the associated cloud costs.

Did you ever see a Distributed Deep-Learning Platform as a Service? Sure not, it’s challenging! Join this session to discover OpenDeepHealth, a PaaS built on top of Kubernetes and designed from principles with a multi-tenancy first approach!

OpenDeepHealth (ODH) is a hybrid HPC/cloud infrastructure designed and developed by the University of Torino in the DeepHealth European project. The goal was to provide a self-service platform for Deep Learning, allowing domain experts to bring their own data and run training and inference workflows in a multi-tenant container-native environment. Kubernetes, the de-facto standard for container orchestration, is the perfect framework for building such a distributed system, optimising resource usage and allowing a horizontal scaling of the infrastructure.

StreamFlow, the ODH workflow engine, can schedule and coordinate different workflow steps on top of a diverse set of execution environments, ranging from single Pods to entire HPC centres. As a result, each step of a complex Data Analysis pipeline can be scheduled on the most efficient infrastructure. At the same time, the underlying run-time layer automatically takes care of workers’ lifecycle, data transfers, and fault-tolerance aspects.

ODH implements a novel form of multi-tenancy called “HPC Secure Multi-Tenancy”, specifically designed to support AI applications on critical data. Thanks to Capsule, the multi-tenant Kubernetes operator, ODH can enforce multi-tenancy at the cluster level, avoiding privilege escalations and exploits, minimising operational costs, and enforcing custom policies to access external HPC facilities.

Finally, ODH provides multi-tenant distributed Jupyter Notebooks as a service through the Dossier platform. This feature gives domain experts a high-level, well-known programming model to write portable and reproducible Deep Learning pipelines, augmenting standard notebooks with resource segregation, data protection and computation offloading capabilities.

I've done my fair share of automating computers and deploying processes. Many things have been said and written and after reading the conference schedule I wasn't sure what to add. Lately, though, I've become intrigued by how computers are automating humans, that is, the creation of photorealistic humans and virtual environments also known as 'synthetic media'.

I will take you on a tour across lipsynching, face swapping, voice cloning and capturing 3d modelling humans. How close are we at generating humans and what does it mean for our society? From Hollywood VFX over virtual production to AI-generated humans via GAN models and deepfakes, I'd like to explain the topics in a technical engineering way to inspire people in this exciting new field.

In order to formalize Distributed Ledger Technologies and their interconnections, a recent line of research work has formulated the notion of Distributed Ledger Object (DLO), which is a concurrent object that maintains a totally ordered sequence of records, abstracting blockchains and distributed ledgers. It has been shown that solving consensus is required to implement a multi-writer DLO. However, it is possible to implement Byzantine-tolerant atomic single-writer DLOs without consensus in a distributed system of n servers, of which f < n / 3 can be Byzantine. 

In the same line, the Distributed Grow-only Set object (DSO) has been proposed, which maintains a set (instead of a sequence) of records in an immutable way: only Add and Get operations are provided. This object is inspired by the Grow-only Set (G-Set) data type which is part of the Conflict-free Replicated Data Types. There are consensus-free Byzantine-tolerant implementations of the DSO that guarantee eventual consistency. 
Fundamental coordination challenges in these objects are the Atomic Appends and Atomic Adds problems, by which multiple records are added to distinct DLOs or DSOs in an atomic way. 
In this talk, we introduce these objects and problems, and we describe how they can be implemented and solved in Byzantine distributed systems. 

The accelerating pace of cloud-native development is enabling faster innovation, but it is also leaving behind an increasing vulnerability backlog. Developers are overwhelmed with vulnerabilities without knowing their actual risk and where to focus remediation efforts. 

Just trying to make sense of the noise already takes precious time away from coding, slowing down the process. Not to mention the frustration of dedicating time to vulnerabilities that don’t matter because they incur no real risk. 

Discover how to filter out the noise with runtime intelligence and quickly prioritise critical issues.

In 2010 the DevOps movement set about solving the blockage on the path to production. The wall of confusion. Developers would throw new software releases, new features and new products over the wall to operations who were tasked with supporting these apps in production. As a community, we've been tearing down the wall of confusion for a decade (sometimes building new and different walls where it stood) and we've learnt a thing or two along the way. In this talk, Hannah will reflect on how we've often "shifted left" the wrong way, and how we might approach this familiar problem in new ways that set our teams up for success. 

APIs are the glue that holds our information systems together. If you run more than a couple of apps, having each of them implement authentication, etc. is going to be an Ops nightmare. You definitely need a central point of management, an API Gateway.

As developers, we live more and more in an interconnected world. Perhaps you’re developing microservices? Maybe you’re exposing your APIs on the web? In all cases, web APIs are the glue that binds our architecture together. In the Java world, we are very fortunate to have a lot of libraries to help us manage related concerns: rate limiting, authentication, service discovery; you name it. Yet, these concerns are cross-cutting. They impact all our applications in the same way. Perhaps libraries are not the optimal way to handle them. API Gateways are a popular and nowadays quite widespread way to move these concerns out of the applications to a central place.

In this talk, I’ll describe in more detail some of these concerns and how you can benefit from an API Gateway. Then, I’ll list some of the available solutions on the market. Finally, I’ll demo APISIX, an Apache-managed project built on top of NGINX that offers quite a few features to help you ease your development.

Within its 8 years of existence, Kubernetes has been the gravitational center of the Cloud Native, elevating a pluggable system that diversified the entire ecosystem. Multiple areas emerged in the industry, galvanizing solutions for components such as network, runtime, storage, and cluster provisioning. The maturity of the cloud native landscape is led by the wider adoption of enterprise and large organizations. However, for these companies deployment and handling of bare-metal infrastructure has always been essential. A pivotal tool to manage cross-provider infrastructure has been Cluster API, leading a unique and radical stance for Kubernetes distribution. In association with a model such as GitOps, Cluster API assembles a mechanism that leverages the concept of a "cluster as a resource.

Music Information Retrieval deals with the development of technologies to facilitate access to large music collections. In this talk, I will provide an overview of data-driven algorithms for the description of music content, with a focus on audio, and how they can be exploited to enhance music listening experiences through visualizations or recommendations. I will comment on the challenges we encounter when dealing with music data, the approaches used to incorporate musical and acoustic knowledge into machine learning models, and the wide possible range of application contexts, from music platforms to personalised music therapy. I will illustrate the talk with a set of demos of software libraries or prototypes I have contributed to in my work.

You don’t always hear architecture and frontend development in the same sentence, however, in modern development cycles, this is changing. Whether you are frontend, backend, or DevOps, this talk is going to give you an inside look at how we incorporated frontend architecture of angular-based apps within a bitcoin treasury management platform that is used by large enterprise businesses. In this talk, Daniel will speak about the technology stack chosen, backend-for-frontend and API gateway patterns, monorepos vs multirepo, platform events consumption by the frontend, and more.

Embedding quality through every phase of the software delivery lifecycle is not easy, but reducing risk and improving application quality is mandatory in a technologically competitive world. How can we improve our CI/CD pipelines in order to achieve this goal? Fran will walk you through different examples of scaling tests early, automation generated directly from requirements, enabling any team to learn from fast and continuous feedback as well as decrease technical debt, and, finally, improve business outcomes by making data-driven decisions about release readiness.

It's 2022. Developers have seen it all. They've tried it all. But it seems all is just not enough. The quest for simple has led us to a path of complexity that is often needless. Or is it? Was it all just a lie? Was it all about job security?

As the world moves towards businesses that demand even more developer resources, can we as an industry come together and think about how we can make the world a simpler place? Or will complexity take over even more?

In Part 2 of Silver Bullet Syndrome, we'll see where we are as an industry, and more importantly, have we finally found the Silver Bullet. 

Observability is the capability to continuously generate and discover actionable insights based on signals from a system under observation, with the goal to influence the system. Stepping back a bit, this is really about data engineering, from pipelines to efficiently storing and retrieving data points to visualizations to applying Machine Learning to automate certain tasks. In this talk we will have a look at the challenges we face in the observability space and how open source and open specifications can help to move the community forward.

The number of BBVA clients who communicate with their financial advisors through our app is increasing by the day, and now exceeds one million text messages per month. In this talk we will explain how we experimented at BBVA AI Factory with generative NLP models to help our managers in their daily interactions with clients. With the aim of building a system that suggests responses to financial managers when a request from the client comes in, we tested a sequence-to-sequence (Seq2Seq) model. As a generative method, one of its challenges is the validation of its outputs. We will share how we tackled this problem in our specific use case along with other lessons learnt throughout the journey that may be helpful for practitioners thinking of using this type of models in real cases.

At Typeform, we believe we can create a world by living our mission: to bring people closer with better conversations, and our clients are at the very center of it. Our customer support agents handle hundreds of support chats and several thousand support tickets each month to help with all the problems our customers have, so we decided to use this direct communication with our customers to deliver the best possible service and value.

In January 2022, we launched Actions After Support, where the overall goal is to understand the relative value of different actions by our customers and identify the best next action for them based on the customer profile and the stage of their customer journey.

We created a recommender model based on customer similarity that is served using a custom UI to our support agents, so they can use this information while having a live chat with our customers.

The model has been written in Python, and it is served using an Airflow and ML Gateway, to a custom UI created on streamlit, allocated as a docker container on a Kubernetes pod. The agents can also send direct feedback about the performance of the model to the data team.

In this talk, I will share with you the whole pipeline used from building the recommender model, serving the results, showing them on a custom UI and the monitoring that is in place, as well as the collaboration between stakeholders and data team. A complete and successful data science use case.
 

The number of BBVA clients who communicate with their financial advisors through our app is increasing by the day, and now exceeds one million text messages per month. In this talk we will explain how we experimented at BBVA AI Factory with generative NLP models to help our managers in their daily interactions with clients. With the aim of building a system that suggests responses to financial managers when a request from the client comes in, we tested a sequence-to-sequence (Seq2Seq) model. As a generative method, one of its challenges is the validation of its outputs. We will share how we tackled this problem in our specific use case along with other lessons learnt throughout the journey that may be helpful for practitioners thinking of using this type of models in real cases.

The cloud journey is a continuous one—there’s no true final destination.

A cloud strategy team defines the goals and roadmap for your cloud journey: Which cloud provider(s) to use (and why)? Which applications move first, which follow in what order? How is success measured along the way?

Migration is the start of every journey but cloud-native is the northern star. 

Follow me in this talk to learn about the best practices and pitfalls of different cloud-native journeys we have travelled with many clients.

This talk presents the essence of what I learnt about fully distributed systems while deploying them in the manufacturing industry, on factory shop floors for the past six years. SQL databases are great resources for local or central data management, but we need something else for collaborative systems or heterogeneous swarms — here I mean the part that is not covered by CRDTs. We also need a different language for live queries against meaningful event streams. I’ll show you what I mean by that.

Many teams are still struggling to implement good APIs, forcing RPC use cases into a semi RESTful world. Modern and efficient IPC is more than just doing REST. Take Kubernetes as example: REST on the outside, gRPC on the inside. We should use this approach for enterprise applications as well.

This session focuses on modern and efficient Inter Process Communication (IPC) for microservices. We start with a REST API, built using JAX-RS and Quarkus to briefly discuss the pros and cons of this approach. Then, we will extend the API with an efficient Protobuf payload representation in order to finally transform the API into a fully fledged high-performance gRPC interface definition. But that’s not all! To put some extra icing on the cake, this talk will demonstrate how to consume the gRPC service from a JavaScript web client and also how to completely generate a matching REST API from an enhanced gRPC interface definition to ensure full interoperability in a microservice architecture.

Synchronising state in a distributed system is a difficult problem. At Permutive, we write software that runs on more than a billion devices every month, scattered across the world. All these devices act as nodes of a giant distributed system, in which state needs to be updated and kept in sync constantly. In this talk, we’ll learn about Conflict-free Replicated Data Types, CRDTs, data structures that can be used to replicate state in a distributed system and allow independent modifications and strong eventual consistency guaranteed by sound mathematical rules. Despite their scary name, we’ll see how CRDTs are relatively simple structures, and how the basic properties they exhibit lead to simpler, safer, and more reliable software.

The Suitcase project is made of a battery-powered suitcase using development boards based on ESP32 microcontrollers Raspberry Pis and a 4G connection and OCI Services. The goal of this project is to enable design and rapid prototyping of stream analysis of events generated by the Suitcase. You will learn how to develop and deploy analytics pipelines that process the events coming from Suitcase and send back the results in case of anomalies and errors.

How do we, at The Hotels Network, build a data platform that helps the hotel industry to attract, engage and convert visitors throughout the online booking journey by using predictive personalization to offer each guest a unique user experience. How that same data allows the hoteliers to benchmark, for them to compare your performance to the market and competition. We will review the tech stack and processes that allows The Hotels Network to process millions of visitors, tens of thousands hotels' data in a distributed way to deliver the data platform.

Adyen provides payments processing and financial services to many marketplaces like eBay, GoFundMe, or Wix among many others. In this setting, any individual can sign up and start selling or providing services and thus there is a need for strong requirements around behaviour monitoring to prevent illegal or damaging situations.

This talk will take you through our journey when solving for behaviour prediction and monitoring an ever-growing dataset. A journey based on several iterations first on posing the problem, designing the right algorithms and then implementing and deploying a live system that leverages the open source tech stack to juggle real-time events together with numerically-heavy machine-learning-based pattern extraction.

In recent years the software industry has begun to pay attention to the field of Resilience Engineering, a relatively new field that studies complex systems and how they behave under unexpected conditions. One concept that has emerged from Resilience Engineering is Graceful Extensibility. Systems that have a high degree of Graceful Extensibility are not brittle: they do not collapse quickly when unusual demands are made on them, and they can adapt to changing conditions.

Many kinds of natural systems demonstrate graceful extensibility - Dr Richard Cook’s talk on the resiliency of bone is a perfect example. Software systems, unfortunately, are typically much more brittle than organic systems - even the largest organisation experiences outages. There is no silver bullet to address the problem that the sheer complexity of software systems causes. However, there are a number of architectural and organisational patterns that we can use to increase the Graceful Extensibility of our systems: from basic circuitbreaking and ratelimiting to more advanced design concepts such as constant-work systems and cellular architectures, as well as organisational practices that help us to spot the signals that our systems are approaching their limits. This talk will dive into these technical and organisational elements of resiliency in distributed software systems.

We're developers. Many of us have or will have kids. And we want them to become proficient in technology because we know the world is just becoming more and more complex. We also know the risks that technology brings to kids' education, communication, privacy and content exposure. Tech humanize and dehumanize. We, as developers, love tech because we can build new ideas with it, but most people are just users. We want our kids to be more than tech users, to learn technology and science from the lens of curiosity and creativity. In this talk, I'll share my experiences and ideas to use technology as a medium of creation, expression, exploration and learning. We'll fly over multiple media and ways to use technology creatively and as a medium to learn scientific ideas.

May contain unpleasant truths.

In order to support modern systems’ needs, we have to be distributed. But, it is easier said than done. Let’s see together what challenges we have in such systems and strategies we can use to overcome them.

We are all aware of the fallacies of distributed computing. Alongside them, there are other things we should consider as well. At AxonIQ, one of our core products is Axon Server which acts as a Messaging Platform and Event Store. In other words, a system that is by its nature distributed across several machines to be more reliable and scalable.

Let’s see together what it means to build such a system, how to maintain it in (large variety of our customers’) production environments, and lastly, how to offer it as a service in the cloud.

Nowadays GDPR compliance is probably one of the most resource consuming areas in all companies. In this talk, we will explain how to develop a machine learning based system that is able to identify the level of sensitivity of data files. Then, and based on GDPR and POPI criteria, it is possible to detect patterns and classify them. From a technical standpoint the problem could be considered as a three class classification problem (a document could be sensitive, highly sensitive or non sensitive). Different approaches were explored during the PoC, combining Named Entity Recognition models with different Deep Learning classifier architectures, and fine tuned pre-trained models. 

Distributed systems are inherently complex and exposed to failures in their components. While distributed algorithms try to do their best in handling node and network failures, it is often not possible to mask them. In the presence of failures, systems that thrive for high availability need to allow users to always submit operations and observe the local state (ensuring local-first), leading to potential divergence. Designs with Conflict-free Replicated Data Types (CRDTs) have been used extensively to guarantee that it is possible to converge to a state that reflects concurrent operations. Recent proposals show that it is possible to go beyond state convergence and use CRDTs as a basis for providing additional system-wide guarantees on individual data items, non-negative inc/dec counters, and cross-item guarantees such as referential integrity. This talk introduces the main concepts underpinning CRDTs and these recent proposals. 
 

More than ever, dataviz needs to engage broad audiences, evolve beyond conventional chart types, and move beyond the usual infographics. Stefanie will use examples from her practice to explore what a ‘post-infographic’ design approach might look like while sharing tips on how to communicate data with a spirit of openness and experimentation.
 

Providing realtime Machine Learning evaluation with multiple platform support, including Tensorflow, BigML and H2O, integrating tightly with a high data availability engine and doing so in Java provides a multi-faceted engineering challenge. This talk will introduce you to the current state of engineering for ML within Java, the current landscape as well as insights from real world design and implementation. Finally, we will discuss the promising future for Java as a fluent platform for Machine Learning. 

During the presentation, we will predict if NASA has found exoplanets (planets outside the solar system). And, for this, we are going to use the results of the Kepler satellite and also IntegratedML, the Machine Learning module for InterSystems IRIS.

There is no doubt that data is being a catalyst of innovation in many companies. However, despite being presented as the “new oil” some years ago, it is not clear that data by itself can make change happen. Whatever their role be, we agree it helps us set the foundations of a knowledge driven business.  In this talk we talk about our journey at transforming a “traditional” listing website to a company able to innovate through data, an exciting but full of hurdles that many companies are going through now.

In this talk, Micronaut creator Graeme Rocher, will demonstrate how you can quickly build optimized Microservices with Micronaut & GraalVM Native Image. Attendees will learn how the combination of GraalVM Native Image and Micronaut can lead to efficient, highly performance, and optimized applications that can be perfectly deployed to containerized environments like Kubernetes.

In this talk Micronaut creator, Graeme Rocher, will demonstrate how to develop Serverless applications in a completely Cloud agnostic way with Micronaut and deploy them efficiently with GraalVM native image.

Imagine a world where you can access metrics, events, traces and logs in seconds without changing code. Even more, a world where you can run scripts to debug metrics as code. In this session, you will learn about eBPF, a powerful technology with origins in the Linux kernel that holds the potential to fundamentally change how Networking, Observability and Security are delivered.

In this session, we’ll see eBPF monitoring in action applied to the Kafka world as an example of a complex Java application: identify Kafka consumers, producers and brokers, see how they interact with each other and how many resources they consume. We’ll even show how to measure consumer lag without external components. If you want to know what’s next in Java and Kafka observability in Kubernetes, this session is for you.

The Suitcase project is made of a battery-powered suitcase using development boards based on ESP32 microcontrollers Raspberry Pis and a 4G connection and OCI Services. The goal of this project is to enable design and rapid prototyping of stream analysis of events generated by the Suitcase. You will learn how to develop and deploy analytics pipelines that process the events coming from Suitcase and send back the results in case of anomalies and errors.

May contain unpleasant truths.

In order to support modern systems’ needs, we have to be distributed. But, it is easier said than done. Let’s see together what challenges we have in such systems and strategies we can use to overcome them.

We are all aware of the fallacies of distributed computing. Alongside them, there are other things we should consider as well. At AxonIQ, one of our core products is Axon Server which acts as a Messaging Platform and Event Store. In other words, a system that is by its nature distributed across several machines to be more reliable and scalable.

Let’s see together what it means to build such a system, how to maintain it in (large variety of our customers’) production environments, and lastly, how to offer it as a service in the cloud.

BPMN is often seen as boring ‘Business Speak’. In reality, it is just a way to automate things, and since it’s called the Internet of Things why not automate those things too! With an estimated 75% of IoT deployments failing to deliver on their promises (and 30% dying in the proof of concept phase) it’s clearly time to approach IoT deployments differently. It’s time to design, implement, build and deploy IoT projects from a business perspective rather than a technical standpoint.

In this talk I’ll go through a complete IoT solution in 3 iterations to show how Business Process Management platforms can quickly iterate on an IoT solution to deliver maximum benefit. Enough business-speak! I’m going to build a Skittles (candy) dispenser based on IoT and controlled by BPMN, with a little AI thrown in! I’ll run the entire demo live, so if the prospect of watching a demo fail in front of a live audience is what gets you excited, this talk is for you! It’s also for you if you’re struggling to build a business case for your IoT project.

Public, Private, Hybrid…Multi-Cloud?

As the cloud development options and architectures continue to evolve, we’ve finally found ourselves at the next new thing: Multi-Cloud. What is it? Do we really need it? How would you take advantage of a multi-cloud solution?

In this talk, I'd like to describe a few real-world use cases where a multi-cloud solution is currently in production, specifically when it comes to data.

Specifically, we’ll see how multi-cloud helps:

• Data Sovereignty scenarios
• Fulfill varied client cloud requirements
• Developers have more freedom in choosing services
• Achieve greater availability

A demo on how to setup a multi-cloud cluster will also be shown.

The audience will leave with a more confident understanding of real-world, multi-cloud use cases and see how to create their own!

Did you ever see a Distributed Deep-Learning Platform as a Service? Sure not, it’s challenging! Join this session to discover OpenDeepHealth, a PaaS built on top of Kubernetes and designed from principles with a multi-tenancy first approach!

OpenDeepHealth (ODH) is a hybrid HPC/cloud infrastructure designed and developed by the University of Torino in the DeepHealth European project. The goal was to provide a self-service platform for Deep Learning, allowing domain experts to bring their own data and run training and inference workflows in a multi-tenant container-native environment. Kubernetes, the de-facto standard for container orchestration, is the perfect framework for building such a distributed system, optimising resource usage and allowing a horizontal scaling of the infrastructure.

StreamFlow, the ODH workflow engine, can schedule and coordinate different workflow steps on top of a diverse set of execution environments, ranging from single Pods to entire HPC centres. As a result, each step of a complex Data Analysis pipeline can be scheduled on the most efficient infrastructure. At the same time, the underlying run-time layer automatically takes care of workers’ lifecycle, data transfers, and fault-tolerance aspects.

ODH implements a novel form of multi-tenancy called “HPC Secure Multi-Tenancy”, specifically designed to support AI applications on critical data. Thanks to Capsule, the multi-tenant Kubernetes operator, ODH can enforce multi-tenancy at the cluster level, avoiding privilege escalations and exploits, minimising operational costs, and enforcing custom policies to access external HPC facilities.

Finally, ODH provides multi-tenant distributed Jupyter Notebooks as a service through the Dossier platform. This feature gives domain experts a high-level, well-known programming model to write portable and reproducible Deep Learning pipelines, augmenting standard notebooks with resource segregation, data protection and computation offloading capabilities.

Everybody talks about security but no one actually knows how to start let alone solve issues when it is not working. We always fall back on “that” guy. With this presentation, I hope that at the end of the talk I inspired some people to become that guy.

Certificates are everywhere: in between; on the server; and even on your own devices. Even the people who don’t know what a certificate is, still look for the lock in the address bar. And if they don’t, they should. For people working with certificates daily, “Zero Trust” is something that is a standard that has been around since TLS 1.0, not a new buzzword. 

When explaining to someone what that lock in the address bar actually means is something you probably need to be working with certificates daily, just to know what that lock actually portrays. Getting from a simple unsecured HTTP connection to an HTTPS connection using a certificate that is not self-signed used to be a lot harder than it is now. Or is it?

Do you know the difference between TLS and SSL, HTTP and HTTPS or even mTLS. Not to mention async vs sync encryption with RSA, DES? If yes no need to join this talk, all others are invited to explore the wonderful world of security and encryption. This talk takes you down the rabbit hole of encryption and certificates and how they are connected with each other. At the end of the talk, you will understand the terms, CER, DER, CRT, PEM, Certificates, CA, Root CA, TLS, SSL and even mTLS and be able to join the discussion at work related to certificates and a bit more.

We've been conditioned to accept that the software we write only runs if we pay our Amazon bill and that the software we rely on can disappear one day because someone else didn't. We've lowered our expectations, too. Our software can only be so fast when it runs on a computer on the other side of the world. 

It doesn't have to be like this. We can have the benefits of the cloud with fewer disadvantages by thinking local. In this talk we'll look at the research Ink & Switch has been conducting into local-first software: software that prioritizes your experience on your computer. We've combined web technologies, recent breakthroughs in computer science, and peer-to-peer data distribution to show how you can build software with real-time collaboration at every level that never goes offline because it runs on your computer. Best of all, by reducing the incidental complexity in development, we believe we can enable developers everywhere to accomplish more with less.

Application security is complex, in distributed systems it is even harder. Agile environments present several barriers to bypass on the way to creating secure software. Scaling sustainable product security in a large organization sounds like a big challenge!

Join this talk to hear about how security is seamlessly embedded in SDLC at The Workshop, while maintaining reliable security coverage for the products we build.

Conversational Artificial Intelligence (CAI) refers to technologies that, via machine learning (ML) and natural language understanding (NLU) techniques, help imitate human interactions, recognizing speech and text inputs and responding in a conversational way. In this talk, I'll teach you how to train CAI solutions including all the strategies you need to minimise human refinement. You'll learn how to create a state of the art solution that can handle intent classification and entity recognition with multitask architectures and original features. 

The remarkable advances in Data, Analytics and AI have fostered a widespread adoption of Machine Learning (ML) capabilities within many products and services we use and consume everyday, ranging from navigation apps up to health-support systems. Nonetheless, implementing ML solutions is still a complex endeavour, and many organizations are struggling to fully exploit the strategic advantages of AI.

In this talk, we’ll cover the key concepts that are critical to harness Data & AI leveraging what we have learned from implementing complex software and data-heavy developments in multiple projects. We’ll go through the three stages that we consider when dealing with data products, and we’ll also introduce the principles of Machine Learning Ops (MLOps), a concept that has gained momentum over the last months.

Lightning talk Winner