Our client is one of the leading car manufacturers from Europe that produces and sells more than half a million cars per year, and generates annual revenue of over €100 billion (2019). 

We have helped our client optimize manufacturing process using predictive algorithm, resulting in a more cost effective process with less waste.

<h2><strong>Client</strong></h2><p>Our client is one of the leading car manufacturers from Europe that produces and sells more than half a million cars per year, and generates annual revenue of over €100 billion (2019).&nbsp;</p><h3><strong>Business Challenge</strong></h3><p>As a part of the production process, molten aluminium is injected into a die cavity, mounted in a machine, in which it solidifies quickly and forms an engine block. Engineers visually check these engine blocks to verify if they are casted well. This project aimed at developing and productionizing a custom algorithm that will decide whether an engine block will be produced correctly during the casting process in order to avoid further processing and scraps, which would lead to financial savings for the manufacturers.</p><h2>Our solution</h2><h3><strong>Machine Learning</strong></h3><p>During the first set of consultations with our client, we identified two thermal cameras that emit thermal images of each engine block as a data source for the algorithm’s training and inference. However, using only the thermal source hindered the algorithm’s ability to correctly classify casted engine blocks, which we proved by experimenting with Computer Vision model only in the first iteration of development. We advised and identified multiple sensors that emit measurements such as pistons speed, intensification pressures etc., and used them as an additional data source that will enhance our Computer Vision model.</p><p>Our final solution used a two-branch Deep Learning model:</p><ul><li>First branch processed tabular sensor data</li><li>Second branch processed thermal images data - four thermal images per engine block</li></ul><h3><strong>Data Engineering</strong></h3><p>We developed two main data pipelines for training process:&nbsp;</p><ul><li>One for transferring the data from our client’s factory server to their remote, powerful server capable of running complex Deep Learning model</li><li>Second for transforming and normalising the dataset, and feeding it to the model</li></ul><p>For the inference phase, data traverses to the model through the same data pipelines, our algorithm produces predictions and sends them back to the factory’s server.&nbsp;</p><p>Additionally, a cron job runs every 5 days, retrains the algorithm, and discards the outdated data.&nbsp;</p><h3><strong>Service Provided</strong></h3><p>Consultations, Design, Machine Learning, Data Engineering, Testing, Maintenance.</p><h3><strong>Project Scope</strong></h3><p>5 months project delivered by a cross-functional team of a Machine Learning Engineer, Data Engineer, product owner and a scrum master.&nbsp;</p><h3><strong>Results&nbsp;</strong></h3><p>Our solution provided an accuracy comparable to an expert engineer and provided significant time savings in manufacturing process, and financial savings.</p><p>The algorithm and data pipelines have been tested in production and were observed to be reliable, and able to handle the load of the client’s needs.</p>

Enhanced Computer Vision Model

Our client is a leading online marketplace for sitting services, such as childcare, senior care, special needs care, pet care and housekeeping helping more than 30 million families in 20 different countries find local high-quality caregivers best suited to their specific needs.  

<h2><strong>Client</strong></h2><p><br>Our client is a leading online marketplace for sitting services, such as childcare, senior care, special needs care, pet care and housekeeping helping more than 30 million families in 20 different countries find local high-quality caregiversbest suited to their specific needs. Client operates under subscription model offering free basic membership tier providing limited capabilities and premium paid membership tier enabling access to all site features.</p><h2><br><strong>Business Challenge</strong></h2><p><br>Search capability lies at the very heart of our client’s business enabling millions of care seekers to match with appropriate caregivers. Our client presented us with the challenge of significantly improving the existing search function with the aim of decreasing customer churn rate and increasing conversion rate from free to premium tier membership.</p><h3><br><strong>Our Solution</strong></h3><p><br>Based on in-depth examination of the platform and extensive domain knowledge our team decided to tackle enhancement of the search function from two angles:</p><ul><li><i><strong>Search Speed</strong></i>: Based on detailed site analysis it was determined that both dated presentation layer at the front-end of the application as well as legacy search engine implemented on the backend of the platform need to be redesigned. Our team has significantly redesigned presentation layer of the search functionality as well as reengineered search engine by migrating the database service layer from on-prem to cloud-based solution, rewriting the service architecture from monolithic solution to microservice based one and providing migration path from the old solution to the new one.</li><li><i><strong>Search Relevance: </strong></i>One of the most important lessons learned from well-established search engines is that key to business success in the domain is improving relevance of search results while preserving simplicity of user specified search criteria. Our team devised an approach for improving search relevance by <strong>personalizing search results</strong> based on search history, job fulfillment history, previous interactions with other users (in the form of shortlisting, liking and skipping certain profiles) and finally the user set preferences set in the search itself. To transform implicit user data into explicit search criteria to be served to search engine the team utilized sophisticated machine learning algorithms tailored specifically towards the specific needs of the project that was specifically trained based on non-discriminatory limited profiles of users to provide results that were both relevant while still being fair to all participants of the marketplace.</li></ul><p>&nbsp;</p><h2><i><strong>Service Provided</strong></i></h2><p>Design, Development, Testing, DevOps and Maintenance.</p><h3><br><strong>Project Scope</strong></h3><p><strong>&nbsp;</strong>6 months project delivered by cross-functional team of 5 software engineers, technical architect, product owner and scrum master, assisted by a team of QA specialists.</p><h2><br><strong>Results</strong></h2><p><br>Solution provided measurable and reliable reduction of search result loading time (from up to 5 seconds waiting for results to an observable maximum of 200ms), that resulted in a significant increase in customer satisfaction scores based on the customer feedback.<br>Personalization of search results has further driven the job fulfillment rate up, with the increase averaging about 15% depending on the market segment. As a result of higher quality candidate results job poster premium conversion rate has seen a significant increase as well as premium membership retention rate.<br>The results have been tested in production and were observed to be reliable, scalable and highly cost efficient compared to the previous solution deployed by the client.</p>

Case studies

Explore the business value we provided to our clients

Enhanced Computer Vision Model

Sitting services website

Solve your future problems today.

© 2023 Arrowhead development