Reporting Back From The Dlabs.AI Hackathon

September might just go down in DLabs.AI folklore. We ran the Hackuna Matathon 2021: an internal hackathon linked to our team-building meeting.

So — if you want to learn:

  • What we did
  • Where we went
  • What we worked on
  • Who won

You’ll find the answers right here.

This is a lighter read, we hope you enjoy it 🙂

The hackathon

Most of the DLabs.AI team works remotely. While that’s great in many ways, the setup does limit how many contacts we have with one another. In fact, the only chance to see colleagues is often on screen.

Fortunately, some generous souls decided it was time to meet up, and we spent a few sunny September days together. We used the time to run the integration linked to an internal hackathon event, which meant combining business with pleasure.

Reporting Back From The Dlabs.AI Hackathon

Meet Krzysztof: our JavaScript Tech Lead and the brains behind the Hakuna Matathon

 

“I really wanted to see as many of DLabs.AI members as possible and meet them: we haven’t had such an opportunity for a long time. Besides that, I had participated in hackathons before, and I know how many valuable things could be produced during those events.”

Krzysztof, Organizer

The first two days were all about the now-infamous event: the Hackuna Matathon. We split into four teams, each dealing with their own project.

Team 1: It’s Not Rocket Science – WBT Calculator

Team members: Emilia Brzozowska, Marta Sudoł, Kornelia Kobiela, Mateusz Kaczorek

The first project was a barcode scanner for the WBT Calculator. The application can scan any product, find it in a global database, and use the associated information to calculate the WBT and WW factors, so a person with diabetes doesn’t have to do it manually.

The next steps the team wants to take to improve the app are:

  • Integrate a feature to add products that aren’t in the global database
  • Add a user panel with more features (like adjusting the grammage of products)
  • 10g/12g switch for WW calculation

Team 2: Suguard Marines – SuguardScout Chatbot

Team members: Mark Silva, Zuzanna Sieja, Máté Kálmán Nagy

The main goal of the project was to create a chatbot dedicated to parents of children with diabetes. It intends to give reliable and straightforward answers to questions or phrases associated with diabetes.

The application was made using the following tools:

  • Google Cloud Dialogflow (a chatbot engine)
  • Python + FastAPI – REST API backend that connected the client application with the chat engine
  • Typescript + Redux + React + React Native – thanks to its modular architecture, it was possible to create a web and mobile interface at a low cost

Team 3: The Plumbers – AI Platform

Team members: Tomasz Iżycki, Jarosław Piotrowski, Marcin Baczyński, Antoine Naud, Tomasz Maćkowiak

The next project was about deploying a time-series prediction model to a GCP using TFX. To improve the model’s life cycle and fully automate training jobs and future deployments, the team decided to build an ML pipeline based on Tensorflow Extended (TFX).

By combining TFX with Tensorflow, Kubeflow, and Google Vertex AI, the team managed to create a fully automated pipeline, which enabled:

  • Automated processing
    • Data loading
    • Data transforming
    • Model training
    • Model deployment
    • Serving model as REST API
  • Serverless pipeline
  • Generic design for either local or cloud training

The model also delivered several important benefits:

  • Reduced development time
  • Fully automated training and retraining process (number of models can be easily scaled)
  • Resource optimization (serverless computations)
  • Automated deployments

Team 4: Food AI – WBT Calculator

Team members: Filip Skurniak, Wojciech Sójka, Barbara Cieszyńska, Marcin Konieczka

The main goal of the project was to build another new feature for the WBT calculator app. The feature lets users scan packaged products such as juice or chocolate, then see information relating to calories, protein, fat, and carbohydrate content.

This is how to process works:

  1. ML model for detecting nutrient table
  2. OCR
  3. Module extracts nutrient values from raw text
  4. Module wraps everything into an API with a ‘hacky’ frontend

“The hackathon was a great opportunity for us to get to know each other even more, take time to learn new things, and improve our own product. We really enjoyed the professional approach towards this competition: ground rules were set, juries were chosen. Everyone was fully focused on goals, received valuable feedback, and found time to grow together as a team. The work-life (and education) balance was spot on!”

Tomasz, The Plumbers

While working on projects, the team still had time to talk and get to know each other, alongside taking time out to welcome two new employees: Basia and Marcin.

“I liked the family atmosphere and the fact a “zero-waste policy” was in place (that works quite well in practice!). Everyone spoke about their projects with joy, and it was obvious that they enjoy working here. Additionally, the fact that the company originated from the concept of helping people with diabetes is fascinating to me.”

Basia, Food AI

Time for informal team building

Saturday was our time to relax. After a morning of hearing the results (while enjoying a lovely breakfast), we visited a nearby town. We arrived, checked into our cottages, then headed straight to the sea to enjoy the sun and relish some time away from the screen.

.

 

Then the evening rolled around, giving us a moment to warm ourselves once more — this time beside a roaring fire — before the main event began…

The 1st Hackuna Matathon Award Ceremony

In 3rd Place…

Ex aequo Suguard Marines and Food AI

In 2nd Place…

The Plumbers and their AI Platform

The Winners!

It’s Not a Rocket Science – inventors of WBT Calculator

 

‘And what of the prizes?’ you ask.

Well, the organizers certainly made sure we enjoyed the sweet scent of success. But besides drowning us in sweet treats, they also gave free access to hundreds of online courses and streaming platforms, helping the team further hone their skills.

Besides that, we’ve been given the time to develop our projects in the future, hopefully one day turning them into an all-singing, all-dancing product.

“The hackathon was a great opportunity to meet people for the first time in person and learn a lot, all while solving a business case relevant to people with diabetes. I’m super proud of the team’s progress because it was the very first day of work for most of them. The setup of accounts and environments was happening live during the hackathon.”

Filip, Food AI

 

“The hackathon was two days when everyone could find something for themselves – both backend, frontend, DS, and non-technical people. Lots of cool topics – a good time to integrate with people and tools you don’t work with daily. I liked that organizers considered people’s preferences (regardless of whether someone is not eating something because of illness or by choice). I was very motivated to do new cool things! Probably the best memory is that the whole event made us do nice things and increased our motivation to do it. I wish there were more such things!”

Marta, It’s Not a Rocket Science

Once the awards ceremony ended, we took more time to chat. And before we knew it, Sunday had arrived. Time to bid our farewell and return home — and despite the sleepiness, we were all delighted to have spent these four days together.

“The renovation of the office itself made a huge impression on me; everything looked great. I can assume the hackathon itself was a good one. Many cool projects, most of which we’ll further develop in the future. The teams themselves were well organized and motivated to work. Integration and the trip were a great opportunity to spend time together in a friendly atmosphere and forget about employee-employer relations. Overall, I felt sad about leaving on Sunday.”

Shemmy, Chairman of the Jury

To sum up…

The event offered the perfect opportunity to build a stronger team bond while having heaps of fun. I think it’s safe to say that everyone had a great time, and we all feel so much more connected.

“The hackathon went without a hitch (somewhat to my surprise!). It all worked like clockwork, most of the company was there, we managed to meet people in the teams. As a result, we have so many great new projects to work on. Together with the rest of the organizers, we even managed to overcome some unexpected obstacles. All I can say is: we know what to expect next year!”

Krzysztof, Organizer

Now — on behalf of the DLabs.AI team, I’d like to thank our irreplaceable organizers: Ewa, Gosia, Kasia, and Krzysztof.

What an event you put on…

You absolutely rock!

 

If you enjoyed what you read, why not join us at our next hackathon? We have plenty of career opportunities, so give them a browse and apply to any that pique your interest.

Related articles

Introductory time-series forecasting with torch

This is the first post in a series introducing time-series forecasting with torch. It does assume some prior...

Does GPT-4 Pass the Turing Test?

Large language models (LLMs) such as GPT-4 are considered technological marvels capable of passing the Turing test successfully....