Company Updates

How Akvelon is Putting AI to Work

At Akvelon, we’re all about problem solving. When it comes to our clients, our goal is to find a solution for every problem while using the best technology, AI being one of our favorites. We’ve used AI to create solutions for a wide range of corporations from tech companies to sports teams to international businesses. Here are six projects that showcase Akvelon’s use of AI and Machine Learning to solve problems across industries.


“Don’t Sleep” App

Worried that you or a loved one may fall asleep at the wheel? We have developed an app that helps drivers stay awake with the goal of creating safer driving conditions for all. The “Don’t Sleep” app works through the front camera on the user’s phone and uses facial recognition and AI to determine if the user looks tired or at jeopardy of falling asleep at the wheel. If the app detects signs of fatigue (if the user is yawning, rubbing their eyes, etc.) then an alarm will be set off and the app will recommend that the user takes a break or finds another driver to take over.

The “Don’t Sleep” app can be integrated with other apps, such as Google Maps, and can be accessed by both Android and iOS users. To keep yourself, and everyone else on the road, safe all you have to do is mount your phone onto your dashboard, load the app, and go! Learn more about this app here.

Technology used in making this app: Java, Dalvik VM, Android SDK (Android) and XCode, Mac OS (iOS).



“Meeting Summarizer” Service

Our developers have created a messenger bot for anyone who is tired of taking notes during meetings. The Meeting Summarizer is programmed to take detailed notes of every meeting and then promptly emails out a short summary of the meeting as well as a full transcript of the meeting. Users will be able to pay attention to the conversations and important points in meetings and let the service take care of the notes. The Meeting Summarizer can be added through a conference call or Skype/Slack call, making it the perfect tool for any office! Read more about this project here. 

Technology used in making this bot: 3rd party speech recognition service (Bing or Google) and a proprietary summarization engine based on the ensemble of models and neural networks.



Facial Recognition Security Program

A province in Canada started a program where gambling addicts could voluntarily add themselves to a list stating they would cease gambling in hopes that it would help them control their gambling addictions. The only problem was that there was no way to keep them accountable and ensure the effectiveness of the program. Our team developed and proposed an app that uses facial recognition to detect if someone on that list enters a casino or an establishment where gambling occurs. This way the program would have the tools to keep members of the program accountable and keep track of those who needed a bit more help controlling their addiction.

Technology used in making this app: Microsoft Azure Cognitive Services: Face API; Windows Presentation Foundation; Web cam integration.



Real Estate ML Predictions

After discovering an opportunity to improve the way people search for homes, our team created a service that can predict the market value of a house based on characteristics of the neighborhood and the surrounding area. This service will help future homeowners find the right house for them and help eliminate places that are not the right fit. The service can also show regression and cluster results so users can choose specific categories such as “price and quality”, “neighborhood safety”, etc. Read more about this project here.

Technology used in making this service: Azure ML, Azure SQL, Asp.Net MVC, Unify Website Template, Web API, Bing Maps, Leaflet, Python, SendGrid. and Porto Seguro

A group of Akvelon engineers and data scientists took part in a competition through (one of the leading platforms for predictive modeling and analytic competitions) where they were tasked to help Porto Seguro, a Brazilian auto and home insurance company. Porto Seguro wanted participants to build a model that could predict a driver’s probability of filing an insurance claim in the next year so they could better understand their customers. Akvelon made a successful model, making the top 0.01% and ranking 84th out of over 4,500 teams. Learn more about our team and their experience making this model here.

Technology used to make model: Adaboost, Xgboost, Lightgmb, and Catboost.



ML Financial Market Forecasting

Our team created a simple way to predict financial market forecasting based on the financial market data and actual news sources. We developed a successful product that uses Machine Learning algorithm solutions and UI to enable control for each step of the process. Learn more about the product here.

Technology used in making this machine: Scala, Apache Spark, Spark ML, SBT, Java, Spring Framework, Hibernate, Gradle, PostgreSQL, MongoDB, Docker, Apache Mesos, Rancher, Jenkins, ElasticSearch, Kibana, ReactJS, and Bootstr.


We’re always excited to work with the latest technology, and we’re just getting started with AI and Machine Learning. Keep an eye out for what we’ll create next!

Interested to see what else Akvelon has to offer? Go to our website and see more projects and testimonies from our clients.