A Look Back at 2023
In the past year, our team has navigated countless growth opportunities, learning experiences, and community engagements. We are committed to sharing knowledge and making impactful contributions.
Here’s a glimpse into the technologies, tools, and ideas we explored, documented in our extensive blog posts, podcasts, and active participation in meetups and conferences. It’s been a period of exciting learning and collaboration. Despite challenging times in Israel during the last quarter, we remained resilient and focused on our commitment to knowledge sharing and impactful deliveries.
Software Engineering
How to Manage Schemas and Handle Standardization
Did you know that the concept of micro-services gained popularity only a decade ago? Micro-services architectures have quickly become the standard for developing scalable systems.
However, this adoption came with quite a few challenges. One of them is the enforcement of communication protocols and data standardization.
The way to enforce a communication protocol between micro-services is by using some kind of an agreement, or in other words — a schema.
In his blog post, Tomer Peleg discusses how Riskified manages schemas and handles standardization in an event-driven architecture.
Dockerizing Spark Structured Streaming with Kafka And LocalStack
If you develop Spark apps, you may be familiar with the integration headache involved with schema mismatch or just faulty configuration.
Tal Wanish's blog post features a tutorial where you’ll learn how to aggregate data from Kafka and write it to a S3 delta table. Moreover, you will write a basic integration test to ensure all the moving parts are correctly wired up.
After completing this tutorial, you will have a local development environment and a test you can run within your CI pipeline or on your local machine.
KISS — Keep It Simple (to) Scale (it)
As software developers, we often insist on using the most cutting-edge technologies, building scalable systems, and creating robust infrastructures. However, as a result, when changes have to be made to the running system it may cause unexpected challenges in stability and overall system operational cost.
Anna Chen, Team Lead at Riskified, faced such a scenario when a small configuration change was requested. Recognizing that implementing this change could adversely affect system stability and lead to high operational expenses, Anna’s team sought a more cost-efficient solution. In her talk at the Women on Stage conference, Anna shares their experience of transitioning from Kafka Streams to Redis. This strategic move not only simplified their lives but also proved to be a more economical solution.
Help Your Org to Make Great Decisions Quickly by Providing High-Fidelity Data
In his talk at the hayaData 2023 conference, Alik Berezovsky delves into the critical components of a healthy data journey for big data and machine learning systems. Learn about the importance of effective data management, from collection to transformation, and discover key steps to ensure the reliability and value of your systems.
Reducing Cloud Costs
Cloud services can be one of the most significant expenses for any software company. But how can a company involve its developers in finding a solution?
In Osim Tochna’s podcast, Guy Machlev, VP of R&D — Platform at Riskified, discusses “Quality Week.” During this week, the entire development team worked to optimize cloud expenses. What is the preparation for such an intense week, and why is it important to set goals for it?
DLQ With Spark Streaming and Avro
In his blog post, Yigal Kassel describes his quest for spark-structured streams that are (almost) unbreakable. He shares how he and his team suffered from failing spark streams due to malformed records, leading to downtime and a lot of pressure. And, of course, their efforts to handle it in runtime.
AI-Powered Decision Making: Automating ML Recommendations in a Critical System
Automation and machine learning have become buzzwords in the tech industry, but they offer more than just hype. We can automate tasks to make them faster, more efficient, and more accurate, but what are the challenges?
Oryan Fuchs shares her thoughts on applying machine learning recommendations automatically in a critical system in her blog post. She focuses on her and her team’s challenges and how they approached them.
DBT as a Self-Service for Data Lake Gold Layer
As a component of our Medallion architecture, we aimed to develop a self-service tool with deploying and testing pipelines that allow our different departments to generate tailored Gold layer tables for their specific needs. To do so, we decided to use DBT (Data Build Tool), a powerful tool for managing data transformations in modern data architectures.
In this talk, Tal Peretz showed how, by using DBT with Databricks, you can easily create and manage data pipelines, transformations, and workflows in a scalable and efficient way. By the end of the talk, you will know how to quickly and easily build a high-quality Gold layer that is optimized for performance and accuracy, giving you the confidence to make informed decisions based on reliable data.
Enterprise Grade Micro Frontends: Practical Guide
Scaling your company’s frontend from a monolith to micro frontends is a challenge. In this talk, Aner Cohen & Yacov Barboi share practical insights on managing consistent states, building resilient applications, and addressing overlooked real-life use cases.
Learn about creating an efficient dev environment, handling multiple routers, utilizing Shadow DOM for style encapsulation, CI/CD techniques, and maximizing TypeScript for type safety. Gain a practical understanding of micro frontends in an enterprise context, whether you’re scaling or planning for the future.
How We Improved Our Monitoring Stack With Only a Few Small Changes
As there are many ways to create a good monitoring system, the importance of a good monitoring system is not questionable.
Aviv Jacobs describes the process of improving the Riskified monitoring system, the questions raised, and the advantages and disadvantages of each approach in her blog post.
Data Science
Data Maps — Creating a Better Model Is Easy if You ‘Just’ Use the Right Data
At the MDLIops 2023 conference, Racheli Abo delved into the complexities of gathering data for ML model training. Exploring the varying impacts of individual data samples on the learning process, she shared techniques to measure their contribution automatically.
ML Evolution: Driving Business Results with Machine Learning
The Machine Learning ecosystem has completely changed over the years, from the challenge of training a single model to the ability to train hundreds or even thousands of models with ease. But, what does this technological evolution mean for businesses? How can MLOps and continuous training bring real value to companies?
In this talk, Andres Asaravicius explored the major challenges of ML over the years and how technology has advanced to solve them.
Build Your Own Lego Model: Connecting Historical Data to Improve Models’ Accuracy
Drawing inspiration from her 5-year-old son’s passion for Lego, Tal Laron delved into the intricate world of data science. Exploring various ways to connect Lego bricks, she drew parallels to the intricate task of connecting and combining different data points in the realm of machine learning, a core focus at Riskified.
The talk showcased practical applications, such as connecting addresses through geographical distance or textual similarity, presenting a hybrid approach that enhances model robustness. Attendees gained insights into leveraging hybrid techniques for improved predictive model performance, leaving them equipped to construct their own hybrid Lego models.
AI: Tomorrow’s Potential or a Risk Surpassing Humanity?
How far is artificial intelligence from one that surpasses human capabilities?
In the D&B DUNS 100 podcast with Hila Peled, VP of Data Science at Riskified, talks about how close we are to such a scenario, whether there’s cause for concern, and how artificial intelligence already affects us.
How Fraud Impacts Marginalized Communities
We held a Biases in Technology meetup, led by Riski-Pride & LGBTech community at our Tel Aviv offices. As part of her presentation, Reka Eszter Bodo discussed how fraud impacts marginalized communities, how to address biases and minorities in your dataset, and how to meet business interests while meeting social responsibilities.
Unlocking Insights: Estimating Causal Effect Using Propensity Score Matching
In his blog post, Nir Shlomo dives deep into the world of estimating causal effects using Propensity Score Matching (PS Matching) with retrospective data. Learn how to balance data, deal with confounding variables, and understand causal inference without spending a lot of money on experiments.
Nir’s guide provides real-life examples, code snippets, and invaluable insights that will empower your research and analytical endeavors.
The Significance of A/B Testing and Power Analysis in Fraud Detection
As part of her blog post, Vered Shapovalov presents possible approaches for collecting data to compare models and discuss their advantages and disadvantages in the context of fraud detection.
She shows that A/B testing has powerful advantages over other options and presents a method called power analysis that can be used as a foundation for a methodology to choose the best candidates for A/B testing.
BI & Data Analytics
Building a Staging Environment for Data Teams
Using a staging environment makes a significant impact on your data deliverables. It improves collaboration with your stakeholders, optimizes your work processes, and assures data trust. Although implementing it requires some work, it’s worth it!
Orit Mansur describes how she and her team developed a staging environment for data teams in her blog post.
In a World of Machines, What Am I? The Evolution of the Analysts’ Role
In the age of real-time performance, analysts play a pivotal role. While manual tasks transition to data engineers or algorithms, analysts are crucial for understanding granular behaviors, spotting trends, and predicting outcomes.
Using his blog, Nimrod Dvir explores the role of analysts, the tech world’s expectations, and the skills needed to optimize models.
What Is Data Integrity, How Do You Achieve It, and Why Is It Important?
In the last few years, Data Integrity roles have grown rapidly, and they function differently in different companies.
In his blog post, Idan Shafat discusses why Data Integrity is important, how to achieve it in your organization, and some insights from his daily routine.
From a Star Athlete to a Data Integration Engineer
How do you make the shift from a “traditional” industry to the high-tech world?
In the Entry Level podcast episode, Idan Shafat discusses his fascinating journey to get to his current role, and how he became a mentor who guides the next generation of analysts, in the same program where he was qualified to get there.
Always Learning
Build the Brand
We celebrated the completion of the third cycle of the Build the Brand course with a special graduation event. At this event, we heard interesting and fascinating lighting talks by the course graduates.
This one-of-a-kind course provided them with all the skills needed to excel in the domain of personal branding, such as writing technical blogs, developing content strategies, presenting to an audience, and much more.
My 3 Favorite Onboarding Practices That Made a Difference
Transitioning into a new tech role can be a challenging task, but it doesn’t have to be!
In her blog post, Shahar Avigezer shares three essential practices she discovered during the onboarding process at Riskified that will boost productivity and help new employees learn how to work efficiently.
Guilds
Our R&D guilds had an amazing year of learning and collaboration! But, what exactly is a guild? A guild is a community of like-minded individuals with shared interests and goals, driven by the mutual desire to improve together.
Throughout the year, our Data, Frontend and Backend guilds organized diverse events featuring insightful talks by both Riskified team members and external speakers. Topics ranged from exploring the React DevTools plugin for real applications to utilizing the Playwright automation framework for front-end product testing, understanding the workings of gRPC and its benefits, delving into Scala data modeling, and much more! It’s been an exciting journey of knowledge-sharing and improvement.
hayaData 2023
Last May, we welcomed over 700 data community members to the hayaData conference, and here’s a sneak peek of what we had!
If by any chance you missed it, want to catch up with our amazing talks, or want to watch it again — here you go — Our YouTube channel features all of our speaker’s talks.
Here’s another chance to thank all the speakers who delivered excellent and fascinating lectures, as well as the sponsors who made this conference possible.
See you at hayaData2024!
Women in Tech
Mommies on Stage
We hosted an in-person meetup for women in R&D on maternity leave with Women on Stage. The very own Merv Dagai, Director of R&D, spoke at this event about how she coped with guilt over not spending enough time with her children, how she moved her family twice for work relocations, how she was able to embrace these feelings and thoughts as she went, and what it means to be a successful career woman and a successful mother at the same time.
Women on Stage Conference
We had the honor of hosting and sponsoring the Women on Stage 2023 conference as part of International Women’s Day!
This one-of-a-kind event featured Technical Speakers Academy graduates presenting their ignite talks, which they developed during the course. This course is designed to empower women, boost their confidence, and launch their speaking careers by giving them the tools to tell their stories. A world-class professional speaker led the course, followed by small groups of mentoring sessions.
Among the Technical Speakers Academy graduates, our own Tal Laron and Anna Chen presented their lightning talks.
Our goal is to continue supporting events and initiatives to help women in tech share their knowledge and experience, hopefully inspiring more women and young girls to pursue careers in tech.
Imposter Syndrome
Imposter Syndrome is a psychological pattern in which an individual doubts their skills and accomplishments, and has a persistent internalized fear of being exposed as a “fraud”.
Sounds familiar?
During her workshop, Stav Levinger, a Software Developer, raised awareness for the Imposter Syndrome, discussed whether or not women have a higher chance of suffering from it, and shared some practical tips for dealing with it daily in our work, to improve our self-worth perceptions with or without external feedback.
The workshop was held as part of an experiential meetup that MSD Animal Health Intelligence Technology Labs created together with the Women in Hardware Israel community.
Giving Back to the Community
Michal Sela Foundation Safe Technologies for Women Hackathon
We were proud to sponsor the Michal Sela Foundation Safe Technologies for Women Hackathon and our very own Hilla Peled volunteered her expertise as a judge in the hackathon.
The purpose of the hackathon was SAFETECH technologies to prevent femicide and domestic violence through the planning, construction, and development of life-saving technological ventures.
Empowering Future Innovators
We hosted grade 12 students from Rishon L’Zion studying cyber & computer science at our offices as part of our CSR program, Riskified Cares!
Shahar Dan began the visit by introducing Riskified and explaining the process behind building our machine-learning models.
Afterward, Nick Grigoriev, Ron Klein, Adi Gal, and Shahar Avigezer led breakout sessions where the students got hands-on experience analyzing orders and even solving production issues.
Our SVP of Engineering, Shai Peretz, rounded out the day with students asking insightful questions.
We can’t wait to see where this young and ambitious group of students will go in the future!
