©2018 by Olmait

Big Data

Innovative, manageable, and cost-effective

For every need, find the right developer. Select the right team of experts, well-versed in R, NoSQL, Hadoop, HBase, Spark, Scrapy and or whatever it is you need. Whether you need a senior expert or a junior starter, with Olmait you’ll find the right fit.

Our pool of talents will get you the competitive edge you need at an affordable price.


Success Stories

Insurance Company Sales Chatbot

Created and designed a question-answer system to sell products to an insurance company’s customers. Worked on intent detection and trained a custom model to develop an incredibly smart chatbot storing statistics on each user. Used the intend detection algorithm implementation, Python, Flask, MongoDB and - last but not least - React for the admin panel.

Face Similarity Check for a Bank

Using Microsoft’s one million celebrity face open dataset to train the Neural Network, creating a full-proof face recognition and user identification search. The face identification software created has more than 90% accuracy. Used Convolutional neural networks, Recurrent Neural Networks, trained a set preparation from MS-Celeb-1M, worked on Neural Networks - PyTorch, and Computer Vision - OpenCV.

eFlow Document Management System for the Government of Georgia

Created a document management system running currently in most Government agencies in Georgia. It has almost 100,000 active users and contains more than 1b documents. The solution was a necessary means to improve and streamline the effectiveness of the agencies, saving the government a significant amount of resources due to easy and fast document exchange. Created the core of the system and its modules using Java, Java EE, netty, threads; Spring Framework, Spring boot; Hibernate ORM; Hadoop, HBase, Spark; Apache Tomcat, Jboss, NGinx; Maven, Jenkins, Jira, Git, Nexus, Mercurial; Solr, ElasticSearch, MongoDB, Redis; and Oracle, MySql, PostgreSQL.

Car Image Classification for an Insurance Company

Developed a neural network to predict car damage from an image, what the car make is, and determine from which point of view the picture was taken. By collecting a major training set and label it correctly, created a state-of-the-art model for predictions. The insurance company then used the Neural Network model to create an app for its users, enabling them to insure their cars without any help of an insurance agent. Used Convolutional neural networks, trained a set preparation from various sources, and worked with Tensorflow, OpenCV, Scikit-learn, and Scrapy.

Contact us

3 Aluf Kalman Magen (WeWork Sarona), Tel-Aviv, Israel

45 Derech Ha'atzmaut (WeWork Downtown Haifa), Haifa, Israel

1/27 Adam Mitskevichi str. Tbilisi, Georgia​