Data Science Lightning Talks 03

Data Science Lightning talks are back!

We'll continue with the Lightning talks series in April. As usual, the meetup will consists of 6-8 very short talks covering various data science topics. The idea of lightning talks is to give more speakers an opportunity to present and to get a better overview what kind of data science is getting practiced.

We have more slots available. This is an opportunity to present what you're working on in your company, what was the topic you studied in your thesis, a tool that you use and want to share, or maybe just a hobby project you deeply care about. In terms of content in needs to be something about data science, e.g. data collection, data pipelines, machine learning models, machine learning in production, explainable AI, data visualization, etc.

Submit your talk proposal here by March 26, 2022: https://forms.gle/G3QAArZi4TJTfGSn7

Presentation rules:

• The subject should be something about data science
• Total length: 6-8 min
• Time for questions after all the talks

Agenda:
17:30 - 17:35 - Intro
17:35 - 18:35 - Lightning talks
18:35 - 19:30 - Discussion & refreshments

Talks:

1. Dejan Petelin (Co-Founder @ ParetoAI, ex Director of Data @ Gousto)
What makes or breaks a data science project
Data Science can surely have a huge impact on business and society, but why do most data science projects still fail? This will be a quick overview of some of my failures and key learnings on building data science (team) with direct impact on business from zero to its unicorn status.

2. Gregor Pirš (Data Scientist @ Databox)
Automatic density visualization
When working on reports we spend a considerable amount of time on perfecting visualizations of our data. But what if we want automatic visualizations that look reasonably good for any input? We will present the challenges of automating probability density plots and possible solutions.

3. Marko Javornik (CTO @ Longevize)
Unlocking the potential of human digital twin
Technology innovation comes in waves. Decoding deep human systems biology is now becoming more accessible then ever before. It is opening a gigantic new market of preventative data-driven healthcare.

4. Jan Hartman (Data Scientist @ Zemanta)
Scaling TensorFlow to 600 million predictions per second
In this talk we'll present the process of transitioning machine learning models to the TensorFlow (TF) framework at a large scale in an online advertising ecosystem. We'll address the key challenges we faced and describe how we successfully tackled them; notably, implementing the models in TF and serving them efficiently with low latency using various optimization techniques.

5. Frenk Dragar (Data Engineer @ Plume, Student @ UL-FRI/UL-FMF)
SloBench: Slovenian Natural Language Processing Benchmark
SloBench is a natural language processing benchmark and leaderboard for the Slovenian language. It is a web platform that supports the creation and moderation of task-specific leader-boards (Question Answering, Machine Translation, Named Entity Recognition...). Users can submit the results of their models, which get automatically scored and posted on the task's leaderboard. The evaluation is model-agnostic and supports any dependencies that can fit inside a Docker container.

6. Erik Štrumbelj (Associate Professor @ UL-FRI, Freelancer @ Erik Štrumbelj)
Bootstrapping statistical graphics
A statistical graphic is both an image and a function of the data. In principle, we can bootstrap it and combine the results into a single image. This noisy image can be a very effective way of communicating uncertainty.

7. Jure Jeraj (Big Data Specialist & Head of Data Engineering Team @ Result)
Big Data Platforms – enablers for successful Data Science projects
You cannot do any good data project without data. But not just any data; ideally you want all possible data. You need to understand your data, to estimate the quality, and you need an environment to perform all those steps. I will share my experiences about hidden steps and elements that can enable successful Data Science projects.