Signals, Systems and Scams: Deconstructing the Future of Financial Intelligence

Welcome to DSF Sandbox Sessions!

Top tech talks and masterclasses from the best in the industry. Every month, and completely free. The catch? There isn’t one. Just a monthly dose of epic content to power up your data passion!

Event details
📆 Date: Tuesday 14th April 2026
⏰ Time: 6:00 – 9:30 PM BST
💡 Topic: Signals, Systems, and Scams: Deconstructing the Future of Financial Intelligence
🗣️ Speakers: Kristjan Erik Liive, Volodymyr Panov, Baran Koseoglu, Zoltan Szopory, Vera Shishkina, Aaron Wilson, Mohammed Topiwalla

Sign up for this event by clicking here!

As global financial platforms evolve, the “Trust Gap” becomes an engineering challenge. Scaling a product to millions of users requires more than just better models—it requires a fundamental shift from manual heuristics to automated, high-fidelity intelligence.

Join Wise for an exclusive Sandbox Session designed for senior data scientists and engineers. We’re moving beyond the basics of model training to explore the “Last Mile” of production ML: the infrastructure of compliance, the democratisation of optimization, and the transition to foundational representation learning.

Through three practitioner-led deep dives and a collaborative product-data science panel, we will deconstruct how Wise builds resilient financial infrastructure where performance and safety are never a zero-sum game.

What we’ll explore:
• The GenAI Reality Check: Moving LLM automations from “cool demo” to “compliant production” in highly regulated spaces.
• Decoupling Optimization: How we built “Threshold UI” to empower non-experts to tune model performance across hundreds of cohorts without a single code change.
• Beyond Velocity: Replacing hand-crafted features with User Event Transformers to capture the deep behavioral context that traditional tabular data misses.
• The Co-Design Philosophy: Why the future of fraud prevention isn’t just a better algorithm, but a tighter feedback loop between Product and Data Science.

Food, drinks, and networking with fellow practitioners will follow throughout the evening.

Check out the full session details below!

Please note: Due to high demand and limited capacity, tickets for this event will be allocated via a random ballot. Submitting an application does not guarantee entry. Successful applicants will be notified throughout March and early April. If you have not received confirmation by April 9th, this means your ballot application was not selected on this occasion.

Talk 1: Evolution of GenAI: Building Compliant LLM Automations at Wise
Speakers: Kristjan Erik Liive, Senior Data Scientist & Volodymyr Panov, Senior Data Scientist.
Abstract: Evolution of GenAI applications in servicing tasks at Wise. We will share examples, discuss architectures and challenges of building LLM-based automations in compliance-heavy space. We will also present our vision and key opportunity areas in further augmentation and automation.
Key takeaways: Practical examples of scoping and implementing LLM-based solutions.

Talk 2: On the Threshold of Greatness: Democratizing Model Optimization at Wise
Speakers: Baran Köseoğlu, Lead Data Scientist & Zoltán Szopory, Staff Software Engineer.
Abstract: This session will talk about machine learning model threshold optimization across hundreds of customer cohorts. Traditionally, optimizing these thresholds has been a complex, time-consuming task, often relegated to highly technical experts. At Wise, we faced the challenge of managing diverse customer segments, each with unique risk profiles and compliance mandates and as a solution we developed Threshold UI we will talk about more in the session.
Key takeaways: Boost your machine learning model performance without changing any configuration in your training pipeline.

Talk 3: More Than Meets the Eye: Transforming User Events into Deep Context
Speaker: Vera Shishkina, Staff Data Scientist.
Abstract: Moving from purely manual feature engineering to representation learning. This session covers the technical POC of a User Events Transformer – a foundational model designed to produce customer embeddings that augment traditional tabular features. We will discuss the proposed architecture, the challenges of building temporal data pipelines, and our vision for using these embeddings to boost performance in domains like scam prediction.
Key takeaways: How learned embeddings provide deep context that traditional “velocity” features often miss.

Talk 4: Panel Discussion – Mind & Machine: How Product and DS Co-Design the Future of Trust
Panellists: Aaron Wilson, Fraud and Victim Prevention Product Lead & Mohammed Topiwalla, Fraud and Victim Prevention Data Science Lead.
Abstract: Fireside chat to understand how DS and Product work hand in hand to keep wise safe.
Key takeaways: How do you find the middle ground between business growth and risk precision? What is the future of DS in fraud prevention?

Schedule:
6:00 PM – Doors open – networking with food and refreshments
6:45 PM – Intro
6:50 PM – Talk 1 (20 minutes + 5 minute Q&A)
7:15 PM – Talk 2 (20 minutes + 5 minute Q&A)
7:40 PM – Comfort break
7:50 PM – Talk 3 (20 minutes + 5 minute Q&A)
8:15 PM – Talk 4 (30 minutes)
8:45 PM – Networking with refreshments
9:30 PM – Event close

Registering for the event: Click the 'Sign up here!' button on the specific event page following the link below. Once you have completed the registration form, you will be emailed a link to join the webinar. You will also receive a reminder link one week, one day and one hour prior to the event.

PLEASE NOTE: Clicking 'attend' on Meetup does not register you for the event. You will need to register for the event on the link provided below to receive a joining link. If you do not, you will not be able to join the event.

Click here to sign up for this specific event

Please note the time zone when you book this event.