Maxilect

Project:

Building Campaign manager and DMP  for globally active MarTech firm. This is an interactive marketing company from US, focusing on the music industry. It offers sales services and solution-driven strategies. The company provides integrated digital marketing campaigns that deliver substantial and measurable results.

Problem:

Client wanted to solve various business problems:

Increase client satisfaction
Increase engagement
Gain new clients
Integrate with key partner (partner wants to remain nameless at the moment)
Redesign and replace DSP module and improve targeting capabilities
Develop DMP module and enrich it with new features
Improve UI/UX
Navigate GDPR compliance

Solution:

At the outset, the whole project was divided into 2 sub-projects. First of all, the solid and scalable back-end including API for key partner had to be developed. Then, the Campaign Manager module redesign and the DMP module development, from the scratch, could take place. Existing UI/UX was to be simplified and standardized.

Scope of work:  approx. 5 man-months.

Outcome:

API for key partner was developed
New Campaign Manager was implemented
DMP module was developed from scratch
Solution was implemented in compliance with GDPR
Satisfied end users and stakeholders

Programming stack:

Angular 5, TypeScript, HTML5, Sass, Node.js, npm, Angular CLI
OpenJDK 1.8, Kotlin, Spring boot/data/web
PostgreSQL, Redis, MongoDB
Nginx, Tomcat
Jenkins, Docker
AWS (ES2, S3, SES, Route 53)

Problem:

Client wanted to solve various problems:

    Replace existing URL shortener

    Migrate historical data

    Redesign and redevelop authorization/authentication service + support single sign-on

    Redesign and redevelop DMP module and enrich it with new features

    Redesign and replace DSP module

    Improve UI/UX

    GDPR compliance

Solution:

At the outset, proper and effective communication protocols with client’s head of product and lead engineer and as well as additional stakeholders was established. Business objectives, product vision, product scope and effort estimates were clearly defined by the client and subsequently galvanized and confirmed by our engineering team. The whole project was divided into several phases. The dev. team designed and developed a solid back-end and re-designed, simplified and standardized existing UI/UX.

Scope of work:  approx. 7 man-months.

Outcome:

    Improved URL Shortener was implemented

    Authorization/authentication service was developed

    Solution was implemented in compliance with GDPR

    Satisfied end users and stakeholders

Programming stack:

    Angular 5, TypeScript, HTML5, Sass, Node.js, npm, Angular CLI

    OpenJDK 1.8, Kotlin, Spring boot/data/web

    PostgreSQL, Redis, MongoDB

    Nginx, Tomcat

    Jenkins, Docker

    AWS (ES2, S3, SES, Route 53)

Problem: The company’s analysis package needed modernization.

Problem areas included a lack of real-time notifications for system critical deviations and inability to perform periodic performance tests.

Solution: A robust custom Business Intelligence(BI) system was developed and implemented to include monitoring and reporting capabilities.

Reliability and uptime were the top priorities for our client so we implemented a well established Microsoft suite solution: Microsoft Power BI base with Data Warehouse system on Microsoft Azure.

Data from sensors is transmitted in JSON format to the servers located in Microsoft datacenters and processed by Microsoft Azure. An end-to-end chain of compatible reliability.

Outcome:

• Real-time monitoring of sensor data and power consumption was achieved
• Statistical data on events is now collected for the analysis of equipment performance
• Features for real-time error detecting and reporting were implemented as per project scope

Technological stack: Power BI, Azure.

Problem: Our client had 7 web portals with unique content and ~15 mln of unique visitors. A major part of online advertising was sold directly (without Programmatic) while ~20% of the traffic was via Yandex RTB platform. The programmatic solutions offered by such major vendors didn’t suit the client’s needs due to the low effectiveness. The clients themselves knew the names of the advertising agencies who were buying client’s ads via RTB.

Solution: We have created a platform which realized the automated process of selling ads via the Programmatic Direct model.

Current status:

• Integrated with 5+ Demand Side platforms
• Scalable and flexible architecture which allows integration with another Supply Side platforms and Demand Side platforms
• Data storage (trades info for the last 4 months, several TBs)
• Reporting platform and journal services
• Custom data management platform (DMP) block
• Supported specifications: OpenRTB, XML, JSON

Technological stack: J2SE, Servlet API, Spring, TCP sockets, Apache Kafka, HDFS, Hadoop Hive, PostgreSQL, Zabbix.

Problem: The client had a strong digital marketing expertise and planned to extend their data integration providers pool to enrich their DMP. We were contracted to integrate multiple data providers and connect them with the existing business intelligence infrastructure.

Solution: It was a challenge of building data pullers across different APIs and integrate the data into the existing client’s RTB and online advertising ecosystem smoothly. We developed infrastructure which allows our client to batch and stream data from providers and forks it into different intra-systems in a near real-time manner.

Current status:

• Data Integration of 10+ data providers
• Processes terabytes data monthly
• Scalable architecture
• Reporting platform

Technological stack: J2SE, Servlet API, REST API, Apache Kafka, HDFS, Hadoop Hive, PostgreSQL.

Problem: Connecting banks, procurement, and retailers to deliver services that meet the demands of today’s mobile merchants as a single comprehensive solution.

The client also required flexibility in the platform enabling powerful partnering options within the merchant services’ ecosystem. Sensitive data also demands high-level security.

Solution: Develop an agile cloud platform that allows quick implementation of new solutions and addresses new market demands. Advanced cryptographic algorithms ensure peace of mind with end-to-end encryption of all sensitive data.

Outcome:

• In production since March 2017
• Iterative development is underway
• Integration with two North American Banks and cloud accounting provider
• More than 5,000 Canadian merchants of different sizes use this solution
• The platform handles thousands of payment transactions daily

Technological stack: Java 8, IBM Technological stack (Websphere, DB2, IBM MQ), Elasticsearch, Web services.

Problem: Hyperledger Fabric – a software framework for specialized solutions based on blockchain technology. Hyperledger Fabric is created to solve business problems through the construction of private blockchain network between business stakeholders and trusted nodes. The framework has a modular architecture that provides a high degree of confidentiality, flexibility, and scalability. Hyperledger Fabric is promoted as an enterprise solution. In spite of that, the main used language is Golang. However, in this area lead programming language is Java. As a result, the platform faces promotion difficulties due to its inability to be naturally developed in Java. Organizations need to hire specific developers or perform additional training of existing developers.

Solution: Hyperledger Fabric registry and the logic of contracts are implemented using “chaincode” which runs on a distributed network nodes. ChainCode is the main part that has to be written during the development of solutions based on Hyperledger Fabric. Our team implements the possibility of writing chaincode for Hyperledger Fabric on JVM languages.

Results:
Ability to write chaincode on the JVM languages:

1. Reduces the costs of developers search and training.
2. Simplifies the development and maintenance of solutions, as you do not have to use different languages of programming.
3. Results in an increase of the Hyperledger Fabric in enterprise solutions implementation speed.

Technology: Kotlin, Gradle, gRPC, Protobuf

Issue: The client hasn’t covered an important functionality of financial calculations (billing) for mobile operators with autotests.

Solution: We’ve provided a test plan with a full set of test cases and a detailed description of all presets and basic concepts. We have implemented automated tests in a combination of Python, PyTest, Oracle-cx, Allure. We have also included new autotests in the Jenkins continuous integration system.

Results:

• We covered an important functionality of the customer’s system through tests, which helped to improve the quality of the product.
• The tests we have created are scalable and not dependent on the environment. Tests can create and delete data by themselves.

Technologies: Python, PyTest, Oracle-cx, Allure.

Issue:  The client has needed automated test engineers in order to accelerate the development of products and solutions, and also quality control process automation.

Solution:  We have developed a convenient framework with the help of which the manual testers had the opportunity to create autotests. The BDD (Behavior Driven Development) approach has been used. During the project, we have created about 200 autotests and checked more than 15 releases.

Results:

• At the moment our framework is successfully used by the client’s testers
• The client is satisfied with the results of our work

Technologies: Java, Selenide, Cucumber, JDBC, Allure.

Problem: The customer actively used marketing traffic channels to promote their services and goods. Data on channels is presented in Google analytics (GA), Big Query (BQ), and in offline databases without integration with GA. The functionality offered by GA didn’t permit full analysis of the data through the sales channels.

Superimposition of offline channels with online channels produced misleading data sets. For marketing expense optimization, the customer decided to expand and modify its analytical tools to be able to deal with complex data structures – integrating more sophisticated attribution algorithms such as Markov chains was a top priority.

Solution: Create a web solution with rich functionality.

Create an infrastructure for interaction between the BQ and the data warehouse for attribution and implemented mathematical models of channel attribution. In the web interface, the user can choose from a wide range of parameters (time interval, channel aggregation type, type of “multi-channeling”, attribution method, etc.) to build multi-channel report.

The generated report is exported in MS Excel format.

Outcome:

• With the help of the custom reports, the customer can now choose a specific multi-channel attribution model
• Reduced budget for marketing and increased its effectiveness of marketing campaigns
• Optimised return on expenditure due to corrected data flows
• Our client was happy to endorse the quality of our work

Technological stack: Python 2.7, Python 3, Django, PostgreSQL, SQLite, Big Query, Google Analytics.

Problem: In spring 2017 the customer approached us to build a solution for a 21st century challenge. On one hand, unite existing taxi company fleets and, on the other, open the door to drivers ready to supply their own vehicle.

Solution: After consulting with the customer to determine the MVP (Minimum Viable Product), we set up a dynamic web portal integrating search and database functionalities and integrated essential external services (SMS, GA, etc.).

Outcome:

• Successful launch in June 2017
• Eventually, the founder has decided to use this IT solution to launch one more startup called PapaJobs.ru targeted to “blue shirts” job search market.

Technological stack: Python, Django, PostgreSQL.

Problem: The customer approached us in October 2016 in need of a RTB platform.

Solution: We quickly formed a team from our talent pool that had an experience in this subject area. Together with the customer, we documented the MVP (Minimum viable product) and started developing the system.

Outcome:

• After 3 months, the prototype was created
• After 6 months, the MVP-version of the system was released to test commercial viability and integration with several contractors
• A monitoring system was also created to ensure stable operation of the platform under the conditions of the constantly increasing of workload
• Accountability bloc was created
• At this moment, the system is capable of processing in excess of 20,000 requests per second and is scalable to handle more bid requests and bid responses

Technological stack: J2SE, Servlet API, Spring beans, TCP sockets, Syslog tcp client, Kafka producer, RabbitMQ client, PostgreSQL.

Problem: Acumatica sought to optimise their site which provides services in the field of Cloud ERP. An up to date system was desired for tracking users with the ability to build reports based on it, to replace deprecated modules and code.

Solution: Our two developers systematically established the development process, automated assembling and testing issues, closely interacted (and continue to interact) with the team from the United States (Seattle). As a result, we managed to significantly increase the speed of page load. We also added the generic map with the help of ACF, and transferred the generation of the html-code to the template engine.

Outcome:

• Increased flow of customers that come through the website (due to the system stability, speed of operation, etc.)
• Developed custom reports of user activities
• Detailed monitoring and analysis of user behavior on the website
• Continued support: At this stage, we are working on further development and technical support of the portal

Technological stack: PHP, MySQL, Nginx, WordPress, Composer, Gulp, Saas, Bower, Amazon Route 53, git (bitbucket). Integration with Google Analytics API (Reporting API, Embed API, Realtime Reporting API), Hubspot API, Google Maps API.