Flink & Kafka Deep Dive: From Data Streaming to Real-Time AI

As real-time systems become the backbone of modern products, the way we design and operate data infrastructure is evolving fast. In this meetup, we’re bringing together experts from Riskified, Confluent, and Wix to show how Apache Flink and Apache Kafka power the next generation of streaming architectures—from resilient, fault-tolerant pipelines and reusable data products to AI-driven decisioning at massive scale.

Whether you’re a data engineer, platform builder, or AI-curious developer, this evening will give you practical patterns and fresh ways to think about real-time systems.

Agenda:
18:00 - 18:30 - Mingling, drinks, and snacks
18:30 - 19:00 - Supercharge Your FlinkSQL Job: Use a Hybrid API Approach to Achieve Perfection - Gal Krispel, Senior Data Platform Engineer at Riskified
19:00 - 19:30 - From Data Mess to Data Products: Unlocking Real‑Time Value with Data Streaming - Ohad Israeli, Senior Solutions Engineer Team Leader at Confluent
19:30 - 20:15 - AI vs. Karen: Building Real-Time Sentiment Analysis Pipelines at Scale - Josef Goldstein, Head of R&D, Big Data Platform at Wix.com
20:15 - More drinks and mingling

The talks will be delivered in Hebrew.

------------

// Supercharge Your FlinkSQL Job: Use a Hybrid API Approach to Achieve Perfection - Gal Krispel, Senior Data Platform Engineer at Riskified
Apache Flink’s SQL interface almost sounds like a dream: write standard SQL - get a full-blown streaming app. However, dreams and reality rarely align, and FlinkSQL achieved a questionable level of production readiness.
In this session, we’ll explore the known upsides and downsides of FlinkSQL. From the enormous velocity and traction FlinkSQL can create in your organization, to sleepless nights of manual state interventions and unexpected data type inference. The good news is, there are ways to bypass obstacles and tap the most out of FlinkSQL’s benefits by using a Hybrid Flink API approach.
We’ll examine the architecture of a fault-tolerant FlinkSQL job (using DLQ), bypass inherent FlinkSQL’s data-type limitations, and finally, show how we can set up our job to handle a massive load. Finally, we’ll also share step-by-step code examples of how to achieve it.

About the speaker:
Gal Krispel is a Data Platform Engineer at Riskified, specializing in streaming technologies such as Apache Kafka and Apache Flink. He focuses on building scalable, real-time data pipelines that power Riskified’s core products. Gal is particularly interested in making complex data architectures accessible and efficient across the organization. His work spans real-time analytics, event-driven design, and the seamless integration of stream processing into large-scale production systems.
LinkedIn

// From Data Mess to Data Products: Unlocking Real‑Time Value with Data Streaming - Ohad Israeli, Senior Solutions Engineer Team Leader at Confluent
This talk shows how organizations can move from brittle, point‑to‑point integrations to a data streaming “central nervous system” that powers reusable, trustworthy data products across both operational and analytical estates. We’ll cover the Stream–Connect–Govern–Process blueprint, pragmatic pipeline patterns into warehouses like Snowflake, and real‑world use cases from fraud detection to personalization—highlighting how reusable data products slash time‑to‑value and operating cost while improving data quality and freshness. Attendees will leave with a practical framework to standardize schemas, materialize real‑time pipelines, and reuse domain data products to accelerate AI/ML and analytics initiatives.

About the speaker:
Ohad Israeli is a Senior Solutions Engineer Team Leader at Confluent with 20+ years of experience in data architectures. He helps high‑growth tech teams turn streaming data into measurable outcomes using Confluent. His focus is real‑time, event‑driven architectures, faster paths to production, and reliability at scale.
LinkedIn

// AI vs. Karen: Building Real-Time Sentiment Analysis Pipelines at Scale - Josef Goldstein, Head of R&D, Big Data Platform at Wix
We’ve all heard that “the customer is always right". But let’s be honest, some customers are just impossible to please.
What if we could detect them before they turn into ALL-CAPS review bombers, in real-time?
In this talk, we’ll walk through how to build AI-powered streaming pipelines using Apache Kafka and Apache Flink that can detect frustrated users and respond in real time at massive scale.
We’ll explore how to embed AI models directly into data streams for live inference, discuss the trade-offs between using Large Language Models (LLMs) and Small Language Models (SLMs), and tackle the real-world challenges of latency, scale, and cost optimization in streaming AI architectures.
Finally, we’ll demo a live end-to-end example, from data ingestion to AI scoring and visualization, that you can later run on your own setup.

About the speaker:
The data and culture guy. Currently leading the Big Data Platform R&D at Wix.
Yossi has over 15 years of experience building data intensive SaaS applications and growing the awesome teams that make them.
He believes that the secret sauce for creating sustainable complex systems is good design practices, effective communication between people, and automating everything. In his day to day he works tirelessly on building the culture necessary to support these values and propagate them to others, one pep-talk at a time.
LinkedIn