2020 was a tough year. So six weeks ago we launched the first ever “MLOps for Good” virtual hackathon. Its purpose? Fostering projects that positively impacted real-world issues by bringing data science to production with MLRun.
Together with MongoDB, Microsoft and 12 leading ML communities, we set out on a journey. To be honest, even we were surprised by the traction it got. More than 300 participants from around the world created over 30 projects. They tackled important issues ranging from healthcare to making the web a safer place.
We’d like to thank all the participants, who took the time and energy to do good in the world. In addition to the prizes, we’ll be donating $10 for charity for each one of you.
We’re now only a few days before the virtual awards ceremony where the winners will be announced. Before we see you all there, we’d like to highlight eight of the incredible projects that have been submitted.
Project #1: Heart Disease Prediction
The first project in this list is a heart disease predictor app, built by Amit Sharma, an aspiring MLOps engineer. By using Flask, Ansible, Kubernetes, Docker, Crio and Jenkins, he created an application that enables medical personnel to submit information and get a prediction about the patient’s heart health. What a great idea to help with the challenging task of identifying life-threatening diseases! Check it out here.
Project #2: AI Wonder Girls: ICU Ops
This important project was submitted by an all women's team: Aruna Sri Turlapati, Anju Mercian, Melania Abrahamian, Kulsoom Abdullah, Sara El-Ateif, Erum Afzal, Nishrin Kachwala and Rosana de Oliveira Gomes.
Together, they built a product that predicts which COVID-19 patients have diabetes. Diabetic patients tend to be more at risk if they get COVID, so this product helps hospital staff give better care and reduce ICU overload.
The team used NLP datasets to create their prediction models, together with Kuberenetes and additional methods like LGBM. See more here.
Project #3: Debrazio
This product’s focus was on sustainability. By using multiple data sets and MLRun, this environmentally aware team built a solution for waste management. The product detects different types of waste, and helps reuse it various ways, like for compost or recycling. See more here.
Fun fact: these team members found each other online and worked across three different time zones.
Project #4: Deepfake Shield
It takes identical twins to understand the power of switching faces, so it’s no wonder twins are behind this project for detecting deep fakes in images. With various models and MLRun, this product can help prevent fraud, fake news, and other malicious activities. See how.
Project #5: Brain Tumor Detection
Brain tumors are hard to identify, and sometimes when patients discover their headaches are actually cancerous, it could be too late. To help, the Visual Velocity team built an app that enables uploading a brain scan MRI image, and getting a prediction on the spot. In addition to the MLOps tools for the models, they also built an informative website to make the app accessible to users. This is what they built.
Project #6: Suicidal Post Detection
Sometimes people call for help, but no one listens. With this suicidal post detection app, agencies and medical clinics can “hear” these cries for help on social media, and reach out. This project tags textual data that has a potential suicidal thought, according to its severity. By using Heroku, Flask, MongoDB and MLRun, this team is on a mission to save people’s lives. Here’s how it works.
Project #7: COVID-19 Detection through Radiography images of Lung CT
Effectively diagnosing patients with COVID-19 can accelerate the treatment they get and stop the spread of the virus around them. So instead of waiting hours for test results, this app detects COVID-19 patients through their CT lung images by looking for infected tissues. To cut down false positives, they trained their model on four categories: COVID, Pneumonia, Lung Opacity and normal images. See more.
Project # 8: PMT
In Latin America, the Proxy Means Test (PMT) is a model for determining which families are economically vulnerable and need aid. But, results are not always accurate. This team is here to try and help. With their PMT app, they created a model that presents an accurate analysis, based on the data it is presented with. More info here.
Join Us at the Awards Ceremony - July 20
On July 20th we’ll be celebrating all of the wonderful projects and hard work put into this hackathon and handing out the prizes. We have some great speakers lined up, including special guest speakers Tomer Simon, PhD, Chief Scientist Israel R&D Center at Microsoft and Boris Bialek, Global Head Enterprise Modernization at MongoDB. So register here and get your front row seat to the ceremony! We can’t wait to see you there.