3 days, 40+ speakers and over 35 cutting-edge tech talks: these are just a few of the highlights from Delivery Hero’s Global Tech Summit 2022. The annual event is a unique opportunity for everyone in Delivery Hero’s global Tech teams to share knowledge across borders, learn, and connect with colleagues from all around the world.
Leveraging the power of global knowledge was the mission of this year’s Global Tech Summit. Delivery Hero has 7+ tech platforms and operates in 70+ countries globally. Our Tech and Product teams have grown 6x times over the last 4 years, which comes with a growth in experience, knowledge and skills that can be leveraged on a global level.
The Global Tech Summit 2022 took place online from November 22nd to November 24th, accommodating different time zones from Seoul to Buenos Aires. Out of over 100 submissions from our Global Tech Community, 35 talks made it to the final agenda – selected by the conference committee consisting of our most senior engineering leaders. The talks covered a wide range of topics including Engineering & Productivity, Data & Machine Learning, People Topics, Reliability & Security, Backend Microservices & Databases, Innovation & Agility, Mobile & Web and Moonshots.
Each topic was led by a representative of our tech leadership team, who moderated the Q&A sessions after the individual talks and engaged the audience. As a cherry on top of our internal speaker line-up, we invited exceptional guest speakers from leading Tech Companies like AWS, Google, GitHub, Lightstep, Cloudflare and Honeycomb.io to share their knowledge and learnings with our Global Tech Community.
Read on for a chance to experience a sneak peek of this year’s Global Tech Summit, featuring three talks on Mobile App Performance, Software Integrations and Machine Learning.
Mobile App Performance
Pero Smiljkov joined Delivery Hero over 5 years ago. He is a senior software engineering manager for Pandora, one of our eight platforms. In his talk, “Mobile App Performance,” he explains how the development and tracking of key performance indicators for our mobile apps are key to ensuring a smooth customer experience. He further deep dives into the performance and reliability metrics used to track and optimize mobile app performance.
“When an app responds slowly to a user interaction or users need to wait to see relevant content in the mobile apps, it’s an unsatisfactory experience. Defining and monitoring a set of metrics that outline user journey of the app is key to understanding performance pain points on end users.”
Pero Smiljkov
Takeaways:
- Optimizing for performance can be challenging yet fun.
- Clearly defined metrics and reliable observability are key.
- Caching is an old but effective technique. Make sure to define expiration time policies to ensure enough fresh data is displayed to users.
Philosophy, Integration and Circuit Breaker
Meltem Seyhan joined Delivery Hero as Director of Software Development for our Turkey Tech Hub in 2021. In her talk “Philosophy, Integration and Circuit Breaker” she shares how most of the runtime problems in applications are caused by poorly implemented integrations and how the circuit breaker pattern is one solution to the problem.
“I believe it is not possible to develop any software without an integration. As someone who has been working in IT for a quarter of a century, I think the most of the runtime problems are caused by the integrations that are implemented just considering the happy path, not the failures. I have chosen circuit breaker as the youngest of all solutions (Martin Fowler introduced this concept in 2014) and also I find it funny and interesting that it is inspired from electricity.”
Meltem Seyhan
Takeaways:
- Be careful when dealing with integrations. Always think of potential problems first.
- Implementing the happy path of the integration accounts for 10% of the total effort. Don’t underestimate the remaining 90%.
- It is your responsibility to minimize the impact of an external system issue in your system.
Are those fries really missing?
Felipe Alfaro is a data scientist in Glovo’s fintech data risk team. He started working at Glovo in May 2021 and worked on several projects related to fraud, including the prevention of refund abuse, automatic chargeback disputes and fraud in cash orders. In his talk “Are those fries really missing” he explains the model his team developed to prevent refund abuse.
“We created a Machine Learning model that calculates a score for each customer to identify potentially abusive behavior regarding refunds. This score has been in production since early 2022 and it’s being used in our Risk Engine to help with self-service refunds and refunds coming from LiveOps.”
Felipe Alfaro
Takeaways:
- A risk score helps to determine how likely a customer is to be abusive.
- We chose an unsupervised anomaly detection model to identify the most likely abusers.
- Identifying fraudsters among new customers is difficult. We use our customer network spots when a customer is actually new or a fraudster creating a new account to circumvent a ban.
Outlook
The Global Tech Summit 2022 demonstrates the unique power that comes from the diversity, scale and rapid growth of Delivery Hero’s Tech teams. It connects engineers across the globe to learn from each other, boost synergies, aim higher and ultimately shape our vision for the future.
We are looking forward to the Global Tech Summit 2023!
Are you interested in finding out more about what we do and how you can be part of it? Take a look at our open positions here!
