Selection of our Interesting Projects
Below is a list of several sample projects which you could work at. Some of them are already under development, while we missed the other because at that moment there were no engineers with the required skills available. Actually, because you were not there :)
We will keep updating this list, so feel free to revisit it from time to time. If something catches your eye and you would like to take part in one of the projects, please visit our Vacancies page - you may find something suitable for you!
We often can't disclose company name due to non-disclosure agreement, but can describe projects as such. Engineers may find them interesting.
Photo Booth for Cars
Yes, there is such a thing. In America, large used cars dealers require high quality images for successful eBay posts and to shoot them is indeed a task. At the same time, the cost of the photographer’s services is high, so the decision is to make a photo tunnel. The car drives through this tunnel at a speed of 5 km / h. Several cameras are triggered along the way and the photos are automatically uploaded to AWS S3, ready to be posted to different auctions. The solution itself is interesting, close to hardware. The car position tracking is made using computer vision. Flashes and cameras are synchronized through the serial interface, and there is a traffic light, so that the next driver knows when the booth is available.
Energy Trading Platform
Electric energy is the same product, just as everything else. The client builds a platform for market analysis and energy trading in the European market. There are different features: Analysis and trends prediction, automated trading and tools for analysts. In general, it resembles a typical exchange system, yet somewhat specific. The consumption dynamics changes at different times of the day, different states sponsor “green” energy differently and it is difficult to stock the goods.
Platform for Investment into Solar Energy
These are Swedish guys. The solution allows ordinary Europeans to invest into solar energy. At the same time, the company itself operates solar power stations, raises money in the crowd-funding format for the construction of new ones and promises a return of about 6% per annum, which is very good for Europe. The platform itself is a rather specific trading system with limitations: There is no free market for solar panels as such, but investors can sometimes buy and sell a panel. And, by the way, the project is made with gamification: You can follow "your" panel and its statistics of work in real time.
E-Commerce for a Beauty Industry Company
Developing an online store does not seem to be difficult, right? It turns out that there are tons of difficulties and challenges when it comes to a company which is included into the Top Innovative Companies list according to the FastCompany edition. Hairstory creates a premium product - innovative hair care products - and uses a complex business referral model for distribution and promotion, the processes in which we had to automate. Moreover, Hairstory sells its product in 35 countries, and the task is complicated by numerous integrators with delivery and logistics services on three continents, various payment services and tax calculation services available in different markets. A number of integrations with Business Intelligence services for analyzing operating activities and automating marketing campaigns and newsletters sits on top this diversity. Oh, yes, Machine Learning is also there - it was used in developing an interactive widget which, based on reviews, ratings and purchases of other users and your answers to several questions, can recommend which product is right for you. Such a simple online store.
Software for Private Pilots - Analyzing Operation Modes of Reciprocating Engines
Private aviation is a large industry in the US. There are enough people who own airplanes and use them in a very common way, like flying to the sea. The software of our client can do two things - analyze data logs of an engine to predict when a maintenance is needed, and help the aircraft owner to properly negotiate with their service station. The private aviation is still an expensive thing.
Software for Aviators – Navigation and Planning
We help our client with this software product. The latter is something similar to Google Maps designed for commercial airlines and private pilots. You can calculate the flight route through the navigation points, determine optimal fuel load by considering weather conditions as well as submit the flight plan to proper traffic controllers.
Machine Learning Models for Medical Research
The client has several statistical models that allow him to sharply optimize clinical studies. By using a small set of data, they can predict how to change the dose of the drug and where to best direct the study. So far, we are trying to help the client load and transform data, but we hope to help with the math statistics as well.
Studying Environment-Friendly and Correct Investment Portfolio
In 2018, people are starting to worry not only about their investments bringing money, but also about getting money into projects that do not harm the environment. The client requested a project to analyze the investment portfolio (shares of companies) by a complex set of ESG parameters (Environment, Governance and Social). Public companies are analyzed by 450+ parameters, among which, for example, is the percentage of women and men in the management, level of recycling, etc.
Drones and Computer Vision for Quarries
A potential client is engaged in a very interesting business. They have a Computer Vision / AI solution designed for quarries (open-pit mines to excavate minerals from the ground). The operator puts several geotags on the quarry (physical devices), and then launches a drone with a camera. The drone collects elevation marks and shoots pictures. The software then builds a detailed 3D map of the terrain from these elevations, geotags and pictures, so that you can plan further developments on this map or even count drilling and blasting! Unfortunately, we lacked enough good Python engineers to take this project - it looked very interesting.
Chat-bot for Insurance Companies
This solution allows insurance companies integrate a dialogue interface into their web-site. It allows resolving frequently asked user questions. The chat-bot can ask customers for details and approximately evaluate cost of their insurance policy and conditions. It can also promptly answer frequently asked questions.
The insurance companies can use a neat admin interface to edit the dialogue tree and set the chat-bot's mood and character.
It is not a secret that nowadays large companies try tracking our actions in the Net to build their marketing. They use our history, habits and preferences. However, such technology is not common in the "physical" world yet.
The client asked to build a prototype to understand whether modern computer vision technology is ready to track people moving around shopping halls. The tasks were to 1) learn to tell people from one another and 2) track each individual and link them to real-world objects. The client wanted to use this tech to analyze behavior of people visiting shopping malls to build adaptive marketing activities.
Recognizing Board Images
This is one more computer vision project. Since our college and school years, many of us remember that it is not easy to draw something on boards. Formulas tend to be illegible and geometry figures - crooked.
The project tries creating a technology which will convert pictures and signs on the fly from the hand-written form into neat vector graphics. To get a nice circle, just draw something roughly resembling it.
Mass Notification Platform
This is a service for education establishments used to instantly inform students of a critical information through several channels, including phone calls.
Say, you need to call 20,000 parents because of upcoming grave weather conditions, doing it fast (10-15 minutes all in all) and making sure that everyone is actually informed. Here comes technical complexity, since the solution requires a sophisticated infrastructure and application architecture.
Carbon Emission Calculator (open-source)
This application helps you answer the questions how much CO2 was produced when manufacturing either a finished product or the whole building. You only need to specify materials used, place of manufacture and suppliers.
Technically, this project is serious at all levels. First, you need to collect and maintain a database with info about all materials and manufacturers based on the Environmental Product Declaration (EPD). The info is non-standard, so that you need to learn how to parse heterogeneous PDF files, extract and normalize the required data. Second, the gathered data itself is not too helpful. You need to understand which materials the product consist of, drill down and measure CO2 emissions on all levels. Say, if we know what armature was used and who supplied it, we can drill down to its steel manufacturing process, energy consumed, transportation and actual manufacturing of the end product.