Projects

We cover a wide range of data science and software development. We've developed NLP, computer vision, time series, recommendation systems, data-driven business development, big data and cloud solutions over the years. We deliver real market solutions (MLOps). We are open for collaboration, joint product development, seeking funding. We are also looking to hire data scientists and software developers.

Detection of lung cancer using artificial intelligence

Detection of lung cancer from radiological images. Collaboration with research physicians and hospitals.

Keywords: lung cancer segmentation, cancer diagnosis, radiology, computer vision

Smart building project with machine learning algorithm

Today, many well-equipped buildings store various data. Using a machine learning algorithm, we searched for patterns in user activities and increased energy efficiency.

Keywords: smart building, machine learning, time series forecasting, gRPC, docker

Detection of hate speech on the Internet

Online hate speech is extremely dangerous; hatred can even devolve into action, and in some cases, cause serious psychological deformities in those subjected to it. We have created an artificial intelligence-based algorithm that can identify hate speech in Hungarian texts, whether that text comes from social media platforms, news, or blogs.

Keywords: online hate speech detection, machine learning, neural network, NLP (Natural Language Processing)

Customer churn prediction

Gaining a user costs a lot of money, so keeping them is very important. If a company knows, at the appropriate time, which users are likely to defect, they will be able to find these users promptly and offer them a convincing opportunity to stay. The goal, in this case, is not just to create a prioritized list, but to ascertain what factors play a role in churn and the optimal time to intervene.

Keywords: Marketing, churn, machine learning, data visualization

Wind turbine failure forecast

By collecting data from the sensors, we predicted and classified the expected failure of wind turbines.

Keywords: Industry, wind farm, neural network, hyperparameter tuning

Development of a HR recommendation system

For freelancers, our algorithm is able to select the projects that are most attractive and relevant to them. Likewise, we show clients the best freelance and entrepreneurial professionals.

Keywords: NLP (Natural Language Processing), machine learning, neural networks, recommendation systems, REST API

Dynamic pricing

Our customer has a subscription model in which users can select a custom package according to their preferences. Using artificial intelligence, the prices of these packages are calculated proportionately to the expected business benefits of that user.

Keywords: dynamic pricing, machine learning

Determining potential subscribers

It is advantageous for a marketing team to consider: What specific visitors already know about the company? How often do they visit the site? What were their previous activities on the site? By answering these questions, one can utilize targeted marketing to improve their conversion rate. In one of our projects, we searched for and found answers to these questions.

Keywords: marketing, machine learning

Customer referral system

Regardless of the economic climate, every for-profit company wants customers. Our product offers clients new customers in the B2B sector.

Keywords: NLP (Natural Language Processing), machine learning

Database cleaning with machine learning algorithms

Used car dealers often enter incorrect data related to the cars in the databases, such as car type, engine displacement, and the number of doors. It would take years to correct these entries manually. In addition to detecting and correcting erroneous data, we also showed, by data visualization, in which cases and how often the various errors occur.

Keywords: machine learning, data visualization

Development of a semi-automatic system for application submissions

Most applications must be submitted with various accompanying documents. Our algorithm facilitates this work for the applicant by using machine learning algorithms to collect a list of documents to be filled in, even from scanned pdfs. It can even identify individual document templates and automatically fill in certain aspects with the applicant’s details. We not only developed this learning system to address this challenge but also created a web interface to interact with it.

Keywords: NLP (Natural Language Processing), machine learning, neural networks, REST API, web development, UI

Facilitate customer service work

With the help of our algorithms, clients have been able to speed up the work of their customer service staff so much that they can complete their work twice as fast and with greater accuracy, thanks to machine recommendations.

Keywords: NLP (Natural Language Processing), machine learning, neural networks, REST API

Fraud detection

Our client’s problem was that employees took advantage of certain situations by secretly writing false consumption data to the administration, to the company’s detriment. By employing our machine learning algorithm, the client could predict these activities. Keywords: time series forecasting, fraud detection, neural networks

Stock market price forecasting system

We have been working in this field for 10 years, developing automated stock robots as part of a recreational team.

Keywords: stock market, time series forecasting, machine learning, neural networks

Recognize the best scenes from streaming videos

Users of our client, a game development company, often stream on YouTube while they play, usually creating 4-5 hour videos. Our job was to find the best scenes in the video so that our client could create a 2-3 minute highlight video out of them.

Keywords: video, stream, youtube, comment

Article recommendation system

Recommend articles to users based on what the user has already opened, read, or rated.

Keywords: Referral system, machine learning