Sessions

We all use cloud object storage extensively, like Amazon S3, Azure Blob Storage, Google Cloud Storage and others. However, not many utilize its unique capabilities and advanced features, such as versioning, security, life cycle... Throughout my work developing Ceph object storage and supporting large customers, I’ve seen a lot of incorrect and sub-optimal usage. This session will briefly discuss object storage design, highlight valuable features and review practical dos and don’ts.

This is a lightweight, (hopefully) entertaining talk, in which I will present insights and anecdotes from my studies in a laboratory that combines artificial intelligence with human-computer collaborations, and how I got to intentionally stumble into people. This talk will include a call for innovations in the border between humans and computers.

Users want to see their data fast, immediately, now. Whether you’re working with an ORM or writing native SQL, you probably had a “scratching-my-head-trying-to-figure-out-why-this-query-is-so-slow” moment. The team at EverSQL analysed more than 30,000 SQL queries from thousands of different companies, to lay down the most common mistakes developers do when writing SQL queries. This talk will focus on writing SQL the right way, the fast way.

Serverless simplifies development and deployment, but typically also comes with low performance, usability and observability challenges as well as cloud lock-ins. Nuclio, a trending open-source serverless platform addresses first-generation limitations, is 100x faster than AWS Lambda and can be deployed in any cloud, datacenter or on your laptop (with Docker or Kubernetes). Yaron Haviv will go over the drawbacks of first generation serverless architectures and will explain how to overcome them with Nuclio’s unique architecture and features. The session will include a live demo showing how to write and manage a fast serverless app from scratch.

Many times we, as Developers and Product Managers, see features or products that aren't being utilized well, or don't fit in with our product or engineering vision, but we keep them just because they're already there. In this session I will introduce signs to identify when it’s time to let the feature or product go and share best practices on how to "have the talk" with the feature. We will review examples of companies who removed features/products and show how, in many cases, removing features improves your product, your offering and you.

As product managers, our biggest fear is creating a product that nobody needs. So, how to make sure your users will want your product? In the B2C world, you have to be both data-driven and customer-centric. To make your product better, you need to: 1. Collect input from users 2. Understand why users churn or stay 3. Apply that data using MVP and fast iterations We tried at Soluto different ways to collect data in the right time and context. I would like to share with you what we did and learned in the process, show methods to collect user input, what worked and what didn’t. I will also showcase examples of how to use the data and create a product your users will actually use - and love.

Bio-Inspired algorithms such as Genetic Algorithms and Swarm intelligence can be competitive alternative for Training Deep Neural Networks for Reinforcement Learning. My journey in this domain started 10 years ago by building a RL-DNN bot that compete in a 3D racing simulator named Torcs. (demo available on my Youtube channel) - In this talk I'll show a different approach to train a deep neural network that can drive and control a racing car that outperforms a human driver in simulated racing environment - laying the foundations to try and solve any kind of RL problem differently.

Our product wasn't complete without this one feature, so we went ahead and built it. Soon enough we came to realize users didn't use it, and even worse, they were still feeling the pain we planned to solve. We ended up finding ourselves in a situation where we had to rebuild a fresh feature from scratch. This experience thought us some good lessons that would should be heard by anyone building anything.

The internet offers an endless amount of simple online games like solitaire, minesweeper and hearts. However, these games are usually done with terrible UI/UX, lame graphics, full of banners and ads, and (dare I say it), built on FLASH. In this presentation I will talk about how I built a DOOM minesweeper game with React, Redux & Typescript. I will also share my dream to create one open sourced website, which will host many well-created simple games, to serve both as a learning platform for developers and for players to enjoy playing good looking games with a great UI/UX, without a ton of ads, all in one place.

In the last 3 years I've being taking an active part in building communities both on my free time & work. These communities (mostly about tech) made real impact on the industry & people lives: From mapping wild fires to save lives with the Air Force to raising mobile chargers prices in Israel (oops), making it into the MAIN news headline and (almost), and so much more... This is no magic, and there are many efforts behind the scenes. In this lecture I will unveil some of them and explain how we, at Wix, initiated a brand-scale global dev communities effort. I will finish with steps that anyone can do in order to take part in the communities world and enjoy its professional benefits.

After coming to the realization that our backend system cannot scale for much longer and that new features are very difficult to add, we decided to write it from scratch. With hundreds of thousands of users actively engaging our system, we don’t have the privilege to start over and grow slowly. We rewrote the applicative layers while still relying on the old data store and then wrote a new db and app stack layer, replacing the legacy one piece by piece. We implemented a migration system that is *always on*, meaning that every change in the old system makes its way to the new system, making the two systems eventually equivalent. I will discuss the challenges and lessons learned.

As a developer, do you want to be a generalist or specialist? The answer is not the same for everyone. It depends on your traits, the company you work for and even how long you are a software engineer. I was a java BE developer for 15 years and then switched to be a generalist. I did everything from web and mobile to BE, data and ops. Being a generalist does not fit everyone but if it fits you it is a hell of a ride. If you are considering being a generalist this is the talk for you.

About a year a go we decided to move from AngularJS to Angular, after 1.5 weeks of development and refactoring we were live in prod. Result were: 0 bugs, 0 down time and most important 0 time was spent on QA. In the never ending progress of Front-End frameworks, you need to iterate fast without breaking your app. In this presentation I want to show the pipeline we built for our web apps, this pipeline gives us very fast way to reach production with very high confidence that no bugs reached production, We will talk about development, testing, build & deployments and how to combine it all to one bullet proof pipeline.

Monitoring is important - but as your company grows it becomes harder to keep an eye on all the different moving parts. As the times roll by and the company grows new technologies are being added to the stack and it’s essential to make sure those can be monitored reliably out there in the wild. We want to share how we transformed monitoring from a one man job to something every developer and Product Manager cares about and actively participates in. This talk will introduce you to open source tools that can enable such a solution. Walking you through a cultural transformation that makes monitoring accessible to everyone.

Ever wondered what's the biological difference between a short summer crush and a life-long romance? Or why we get tearful at our best friend's wedding? And what does that have to do with how users fall in love with our product? We often hear that we should "get users hooked" on our product. Asking them to love it can be much harder. In this talk, we'll show how the hormones and neurobiology of the brain can make us TRUST and LOVE a product with a meaningful relationship, rather than just get addicted to it. We'll learn how some of the best products in the world use our neurobiology to make us love them, and try to create a useful toolbox of knowledge we can use in our own applications.

Each of you probably quoted Conway's Law, or Murphy’s law more than once during his career. Did you know that there are many other quotes you can and should be using? Quotes and adages are great in capturing insights and defining a common language - kind of like Design patterns are. So, are you going to stick with only Singleton and Factory in your arsenal? Knowing some additional quotes, the ideas and anecdotes behind them can help you harness industry insights in your favor! But not only that, it would also make you look smart and funny along the way! In this talk I will cover some such “laws”, insights they capture, the rational and people behind them.

I've shared more than 2000 posts on SoftwareLeadWeekly.com in the past 5+ years. To share these posts, I had to read almost 14000 posts during that time, which includes blog posts, videos, books and interviews (podcasts). I want to share some of my lessons learned about: - Which kind of "Growth Material" you should focus on? - Where and how to find great content (good value vs. noise ratio)? - How to set time for learning? (I'm a father of 2, I know how hard it can be) - How to consume knowledge X2 faster? Are you seeking ways to up your game? If so, you'll enjoy this talk.

Just joined a company full of super-senior-10x-developers and you feel lost and scared? You want to boost up your career and become a better developer? Or maybe you just want to have more impact in your current role? The talk will focus on the state of mind and actions that can help you to maximize your impact, and improve your software skill level :) It is based on my own experience as a super fresh and inexperienced graduate joining a company where the concept junior developer didn't exist, to a meaningful team member, and later becoming a lead developer. The lecture will tackle a super important question: <b>How can you give and get the most value from your job?</b> come and find out.

While finding good software engineers is hard, finding the right ones for your team is harder. As we grew our engineering teams, we found ourselves putting a lot of time and effort into not just the hiring of employees but also the fine-tuning of the hiring process itself. In this talk, I'm going to share with you how we use games as part of our interview process. What led us to use this technique, a few takeaways from running this experiment for the past year, how you can use it today and of course - a live demo where we defuse a bomb together

In the last decade, the centrality of front-end components vastly growing. Developing front-end in a large scale requires declarations and enforcement of principles and guidelines for keeping the code maintainable, extensible and stable. My talk main notion is to introduce the S.O.L.I.D principles within the context of front-end components. I will discuss the reasons for implementing the S.O.L.I.D principles in our renew web applications as a comparison to our starting point. I will suggest guidelines based on my interpretation of S.O.L.I.D in the front-end world. I will enrich the discussion by supplying various practical examples regarding the guidelines implementation.

One DB to rule them all? Hardly. Typical applications start their life with one DB, usually the one most fashionable, reliable, well-documented, referenced-on-google, or all of the above. Soon enough you find out not only that that DB you chose early on is actually wrong for you, but that it’s really hard to get rid of. In this talk you’ll see how in under two years, our application turned from single-DB, small & tidy, into a monster running four+ databases (on a good day), while testing and abandoning more than a few others on the way. I’ll discuss lessons learned the hard way on the process - about choosing a database, writing good code, making good products, and life in general.

REST is commonly used and considered as the `usual suspect` to solve any inter-process communication. RPC, data dump, message notifications, actors model, are all valid ways to communicate between distributed systems. In this talk, I will explore the different approaches and technologies in order to help you pick the right architecture for your domain.

During the Cold War, the CIA knew how to expose Russian spies disguised as American citizens with 100% certainty. They used only a piece of paper and a few questions. How did they do that? Hacking your mind is easier than you think. Let’s explore how these mental hacks affect the code we all read and write. We’ll take a stroll through the world of cognitive psychology, and shed some light on some of our industry’s best and worst practices. We’ll have a few interactive examples of our mind’s limitations, examine how these limitations manifest themselves in real code samples and engineering practices, and take away scientifically backed techniques on how to write better code.

In 2008 the market collapsed worldwide and our one-year-old infant startup was close to joining the sad statistics. We almost run out of our seed funding and our product results were poor. After cutting down salaries we were able to get some additional funding to run the company for few more months. We understood that things need to change in how we build our product and how we validate our offering. 6 years later, we sold Check to Intuit for $360M. We are now running Next Insurance on the same concepts with exceptional results. In this talk I’ll explore how we put lean startup concepts hard at work and how we’ve been able to take our product into massive acceptance by millions of users.

As event sourced micro-services become one of the most popular architectures to build a distributed, data consistent, large scale systems, many companies find themselves facing the ultimate equation of System Efficiency vs Cost. Being ready for large scale data inflow at any given time while keeping the compute cost at its lowest is a challenge we have faced at Trax as well. In this talk we will present our dedicated solution using some of the basic APIs given by all cloud provider. We have ended the journey by both reducing the delays in the system to minimum and by saving ~65% of the compute cost.

Telling a cat from a bird? that's easy, most infants can do that. But how about learning to paint a black and white photo with real color? It's time for your models to grow up. We find that transfer learning from different datasets and tasks saves a lot of time and money when labels are scarce and data is limited. In this lecture, I review self-supervised methods that are used to pretrain models on unlabeled data. Methods from the fields of Vision, Audio and NLP will be refined so they are applicable with other domains and effective on your data.

Open source software is one of the biggest game-changers in our world. It has a lot of benefits, the most critical one being minimizing the time to market. But it’s not a free meal, and you should take care of many aspects when choosing one: * Bug fixes (“wait a minute, I need to know $lang?!”) * Security (“who the f*** gave full access to my credentials dir”) * Stability in different bad cases (“why did jobs execute dozens of times when workers lost connection”) Those can catch you one day in a very unpleasant situation. In this talk I will take you on a tour of some key indicators that you need to bear in mind when choosing an open-source library or component.

Whether we like to admit it or not, we've all borrowed code from stackoverflow at one time in our lives. Many do it more often than they care to admit. If a vulnerability exists in a stackoverflow code snippet, it's easy for it to go viral in even the most widely used frameworks and libraries. In this talk I’ll share the findings of the research we did resulting in several wide reaching vulnerabilities affecting different ecosystems (Java, Node.js, Go, Ruby, .NET) and many of their applications and libraries. We'll look through the technical details of the vulnerabilities, and what can we do to avoid them.

It’s almost midnight. Me and my daughter are looking for jumpers, matrices and resistors. There is a bug in the game we’re building, and it doesn’t seem as if we are going to bed anytime soon. Teaching kids recursion and data structures didn’t create the thrill I thought it would. But building games does! My daughter Roni and I will talk about introducing kids to the world of programing, 3D printing, Scratch, Arduino, Ali Express orders and hot glue. We’ll cover the existing tools from which kids can learn how to code, at what age should they start, and which projects to choose. And why girls at school are anxious about coding? Is it a boy’s club already before high-school? 

In this talk we'll take a hard look at one of the most commonly used, and at least as commonly misunderstood, elements in software engineering: time. Time is so fundamental to the way humans experience reality that we don't normally give it a second thought, but it's just as fundamental to software systems. Without a correct model for working with time BAD THINGS HAPPEN: data is persisted out of order, exceptions occur where they shouldn't be possible, and production systems blow up. We'll cover the various common representations of time, acknowledge their caveats and deficiencies, and hopefully learn a few new tools and practices along the way.

When was the last time you wrote an algorithm to plan your diet? As programmers we do amazing things in our everyday job, but rarely do we use our knowledge at home. In this talk I will introduce genetic algorithms and share how I coded a genetic algorithm from scratch and used it to generate my weekly schedule and to create a smart diet planer. We will go through the different stages of the algorithm and understand how they affect the algorithm’s solutions. Let me show you a different side of genetic algorithms and you will discover a new way to solve your everyday problems. This sessions is presented in English.

Recommender systems need to optimize a delicate balance between exploring new recommendations and exploiting informative ones. Up until this point recommender systems employed mostly bayesian prediction algorithms and utilized the inherent measures of uncertainty to optimize the crucial exploration/exploitation balance. While DNNs obtain state of the art results, they present specific challenges for recommender systems due to their lack of standardized uncertainty measures needed for most exploration strategies. In this talk I’ll cover how we overcome this challenge at Taboola and how we use uncertainty measures to our advantage when serving content recommendations billions of times a day.

Progressive Web Apps are changing the way we think about websites and blurring the lines between Native apps and websites. PWAs enable websites to add features we previously could only use in mobile such as: * Push notifications * Full Screen * Offline working * Splash screen, to give an app like feel * App Icon on the home screen In this talk, we'll walk through all the game-changing features PWA offers by building our own PWA. The goal is to walk away with an understanding of the powerful features PWA's offer. Additionally, we will see firsthand how to convert a site into a PWA that’s just like a native app, but without the excessive file size or the lengthy install process!

Klarna is "like Paypal, but in Europe, stronger on consumer, (probably) breaks more things, has too many official programming languages to count, and is allowed to be on the cloud. It's a tech jungle.". We had to introduce data driven decisions into an engineering process on a micro-service architecture layout. An experimentation platform that can work on N different languages (Java, Erlang and Haskell just to mention a few), has a <1ms overhead per call, no SPOF, extremely resilient, deployed on backends and frontends, had to be invented and then be built. By one person. This is a story about how I built it, and survived.

They are back! After the viral success of last year's interview song (https://www.youtube.com/watch?v=lUubm2B4VIU), Iftach & Yoni return with an epic ballad about an epic programmer. This will be a satirical song, mocking some of the unique patterns of modern software development in Israeli hi-tech. There will be laughter. There will be tears. Don't miss out.

I will tell my story. A story about a dinosaur, Mono in the wild wild front-end. At first look I was naive, I thought I could tackle it with component approach. I realize I got it backwards and my problem was scale so I turned to micro-services. During my work, I noticed that everybody started taking this road, but the results were different and there was 0 collaboration. Recently a new term was born, "micro-frontends", but still the status stayed the same. The story continues with: How I dealt with the unique characteristics of my organization. The problems we encountered and how we overcame them. What I think will be the future and how we can understand and collaborate better

Do you have the dream being an independent and stop being a slave of your boss? Do you have a free moment to be self employed? As a freelancer for a couple of years, I'll explain to you what are the reasons not to do so. Of course I'll also explain the what are good reasons to do it and how to do it right.

Think of the new tools that are taking over the Javascript ecosystem: Babel, Typescript, Rollup, ESLint, etc. What do they all have in common: they all take Javascript source code as input, and some emit Javascript code as output. This talk will be a deep dive into the basic building block all these tools share: Transforming your code into a JS Abstract Syntax Tree (AST). You'll learn to read, traverse, and manipulate the AST so you can extend Babel by writing your own plugins, or by writing custom ESLint rules to enforce your team's code conventions. You'll also learn how to create super powerful "code-mods" to automatically convert thousands of legacy ES5 scripts to ES6 in seconds.

Have you ever been asked to implement some crazy idea? to scratch your right ear with your left hand? There are all kind of situations in which, as a Data Scientist, you may be asked to use your magic and find a solution for a problem that seems out of your scope. Whether those reasons are business constraints, customers demand, unavailable simple solutions, it all comes down to the same question: Would you accept the challenge? In this talk we’ll explore a real-world challenge and try to answer the following questions: - Does a “Data Science” approach make any sense for a certain problem? - What other alternatives are available? - How much time and resources should be spent?

A few month ago, I got my new 3D Printer. As a Kickstarter project, it came as half-baked product - its firmware topped with many annoying bugs. In hope to fix some of them, I went to look into the firmware, but alas - it was encrypted with some sort of substitution cipher. In this talk I will show you how I used some data science, statistics, ARM architecture knowledge and much guesswork to defeat the encryption of the firmware. We will see some Python code and I will walk through some IDA scripts I built especially for this mission. Let the firmware's secrets reveal themselves!

‘When the web came out it CHANGED everything! It took us 20 years to realize that it didn’t change anything! The web has dominated us to think that it’s architecturally significant’ - Bob Martin. Today, many believe that Deep Learning is changing everything - applications don’t have to be explicitly written. But does it? DL projects have many of the same challenges as any other software project while it grows to support more features and use-cases. In parallel, most DL practitioners have little background or passion for Software Engineering. To show the importance of SE for DL - I redesigned one of the most popular DL repos to be at least 10 times easier to read and extend. Join the party!

Distributed systems can seem magical, and sometimes all of the magic works and our job succeeds. However, if you've worked with them for a long enough time you've found a few places where the magic starts to break down and the fact that it's actually a collection of several hundred garden gnomes* rather than a single large garden gnome. This talk will use Apache Spark, Beam, Flink, Kafka, and Map Reduce to explore the world of data parallel distributed systems. We'll start with some happy pieces of magic, like how we can combine different transformations into a single pass over the data, working between different languages, data partitioning, and lambda serialization. After each new piece of magic is introduced we'll look at how it breaks in one (or two) of the systems. Come to be told it's not your fault everything is broken, or if your distributed software still works an exciting preview of everything that's going to go wrong. Don't work with distributed systems? Come to be reassured you've made good life choices.

The everyday goal of developers and DevOps is high quality software delivery in high velocity. Security is not their focus and we believe it should not be a burden as well. Security should be natively and seamlessly integrated into the tools and platforms developers and DevOps use, so it will become effortless. Conjur Open Source is a security solution built with developers and DevOps in mind. It can free the developers to do what they do best – develop great software – but without compromising their security. In this session I will introduce Conjur Open Source, walk through how native and easy security can become and how you can get started and contribute – right after this session.

Whether you are a developer, a team leader or a product manager, your role involves making decisions on a daily basis. For most of us, making those decisions can be hard. This presentation is a cheat-sheet on how to hack your decision process. Using only a few easy steps and a cheap gadget, your decision-making process will become faster and more efficient.

The graphdb world hasn't changed much for the past couple of years, the same major players applying the same old know techniques, when I started creating my own version of a graph database, I followed the same path, but to make a difference and really standout in the crowd something had to be done differently. This talk will introduce you to and guide you through what I've learnt in the past two years creating the fastest graphdb today using basic mathematics, I hope your linear-algebra is polished!

As researchers and developers, we constantly face new problems, and challenge existing solutions to old problems. On beautiful circumstances, solved problems from one domain shed light on new problems from totally different domains. This happened to me while I was researching a Ride Sharing problem of optimally grouping multiple riders on a single taxi - a DNA Sequence Alignment algorithm gave inspiration for a hybrid solution, that turned out to be significantly more efficient than the original one. By sharing with you this story and its resulting algorithm, I hope to entertain your curiosity, spark your creativity, and encourage you to venture into unexpected solution spaces.

It still amazes me to see the basic security flaws over and over again on every site that I check, All of those security breaches can be prevented and closed quite easily. In this talk, I will explain and show basic security flaws in web applications - how to detect those and how to make sure that your application or web site will not have those.

In our industry, there is a clear differentiation between B2C and B2B. And while the business models and sales funnel can be different, we tend to have a different approach in terms of UX and product. But, why? The people that are using our products don't stop being consumers the moment they enter their office, and the high level experience they are expecting from the apps in the their pocket don't disappear when they sit on their desks. In this Lightning talk I am going to share some examples, the lessons I learned from working in a startup that creates medical devices for doctors and nurses all around the world, from Africa to the USA.

More of today's applications are being built from smaller components and modules. However, the workflow around building with smaller components can also generate a lot of overhead. Bit is an open source project which helps developers discover, use and collaborate on shared components while distributing the development process itself. In this talk I will introduce Bit, talk about the challenges and opportunities of building a core piece of your technology with the community, the challenges of having to play hand in hand with the rapidly-changing open source ecosystem (from Git & NPM to Webpack and React) and share some insights for teams who want to open source some of their projects

Most presentations in the software world describes a positive achievement of sorts: a feature added or a bug removed. This presentation is different. In this presentation we shall describe how we deleted over 4000 lines of code, together comprising 30% of the code base of a complex device driver, without even trying. Last year my team started working on refactoring the Arm CryptoCell device driver for the purpose of integrating it upstream. While doing so, a pattern emerged - almost each set of changes would reduce the code line count, and yet the functionality stayed exactly the same. This presentation explains how and why we cut 40% of the code with no functionality loss without trying.

Have you ever received an email from your aunt saying paracetamol is dangerous because it contains a rare virus? Heard in the news that kids don't need antibiotics? Read online that garlic is enough to stop rabies? We are surrounded by bits of information, many of them are false. Some of them are dangerous, and could hurt people's health or even cost lives. In this brief talk we will learn how to identify fake news and how you, too, can help make a healthier society.

Development in some open source ecosystems is very different in nature from regular 9-5 development. We’ll talk about the surprising ways open source is different from regular development, the importance of people, technology and teamwork. We’ll go through real open-source changes affecting millions of developers in Node.js 11 and Node.js 12 and see how the people behind them work and how to get involved yourself.

Reinforcement learning is an area in unsupervised learning that deals with learning out of experience. The machine can be described as an agent, taking actions in an environment and getting feedback, while trying to maximize some reward. Deep learning models achieve very impressive results in reinforcement learning for playing games such as chess or AlphaGo, and recently even won some world class players. In this talk I will cover a deep learning approach to play the 2048 game - using popular techniques such as fully connected and convolutional neural networks. We will cover a simple trick, which as it turns out brings impressive results with very few trials.

At last year's Reversim, Iftach Bar gave a beautiful talk called “Lets talk about your salary”. He taught engineers how to negotiate their salaries when getting hired. But while salary negotiation is hard for candidates, it's practically a nightmare for employers. When we hire we're faced with the impossible task of evaluating the potential value of a perfect stranger to our company, knowing that the cost of being wrong could be extremely high (especially for startups). In this presentation I will share how at BigPanda we try to minimize that risk so that eventually both sides are happy. It'll also give candidates a chance to better understand how managers think when making offers.

I front-end today, we are almost always dealing with abstractions - a framework mediates our interaction with the DOM, a library wraps our events, heck, we even have a library enforcing types. Even though we don’t always see it, underneath it all is still good-old JavaScript and web APIs, and it’s worthwhile to know what our abstractions ultimately translate to. Not only will it make us better developers, not only will it make us better “abstraction-users”, but also - it’s something we are all completely capable of. So let’s dive into two such abstractions - ReactJS’s component lifecycle and Angular’s EventEmitter - and look at the source code and the nuts and bolts that make them possible.

We’ve all heard it. The “business guys” want everything done, working perfectly, yesterday. The devs on the other hand say it’s not ready, any of it, ever. Such extreme stereotypes are rarely accurate, but the inherent business <-> R&D tension is very real. In this talk we’ll discuss the differences in perspective and language between the different roles in a tech company - and understand how can everyone be right while holding opposite opinions. We’ll then go over several tips on how to communicate the dev point of view in the language of business - while making sure everyone are aligned around the greater good of the company.

The computation of large systems' atomistic dynamics is required in fields such as biochemistry, electrochemistry and many others. Fully quantum molecular dynamics is a powerful tool, but can have a high computational cost. An approach that was developed in the last decade is to use ML algorithms to build on the fly computationally cheap predictors for the energy, forces, and other physical properties. This approach enables the performance of calculations with an accuracy that is close enough to fully quantum molecular dynamics but with running speeds that are more than 100 times faster. We describe and analyze the construction and use of a DNN based model for the forces in solids.