Softkit

Project Overview: 

A platform for guiding and teaching your audience across the globe. It allows for creating customized and gamify any course using interactive quizzes, games, and videos. Doesn't matter where you are - the app is always easy to use for studying or teaching.

Provided Services:

Development of adaptive version, subdomains, and custom domains; speed optimization, security settings; implementation of the subscription system, referral system and discount/pone system, Stripe integration

Value:

Previously, it was necessary to restart the project manually for the deployment - now it can be done automatically thanks to Bitbucket Pipelines, which allowed to speed up the development process and save more time.

Challenges:

1. The Letsencrypt certificates generated by Certbot were not used without restarting the front end.
2. The database was subjected to heavy loads and the content on the pages was loaded too long due to the large size of the images, which also took up a lot of space.
3. There was no protection on the site, so it was necessary to take measures to achieve the maximum level of security for the platform
4. Implement a fully customizable landing page for teacher/school/courses.
5. Increase the reliability of the database to avoid losing user data.

Solutions:

1. The problem was solved by delegating certificate generation to an auxiliary server, which now redirects (proxies) requests to the main server.
2. It was decided to transfer all images to the AWS S3 and compress them to the optimal size using a special library. Also, to the site was added a cropper: when uploading a new image, the user can choose only the desired part, instead of loading the entire image.
3. Since everything had to be done from scratch, user authentication was rewritten. JWT tokens were added to authenticate them.

4. A settings page was created in the teacher's account where a user can:
   - upload image for the header, logo, and favicon
   - set up fonts (Google Fonts API was connected for this)
   - choose a color theme: from ready-made assignments or a configurable theme (the ngx-color-picker library was connected to select a color scheme);
   the configuration of the Page Landing Theme is saved to the backend. After on the pages themselves, a request for configuration is made and the inline-styles are already distributed on the HTML elements of the landing.

5. The database used on the project has been moved to Amazon RDS - this service automatically backs up and reverts to earlier states (restoring earlier versions).

Technologies & Tools: 

Languages: Java, TypeScript, Bash, SQL

DataBase: MySql

Tools and frameworks: Spring boot, Spring MVC, Spring JPA, Spring Security, Angular 8, HTML, CSS, Docker, EC2, Amazon S3, Amazon RDS, REST, Maven, Git, Bitbucket, Swagger, Stripe

Project Overview: 

A first-of-its-kind educational resource and proprietary recommendation engine for homeowners researching and evaluating the many ways to sell a home.

The recommendation engine is powered by machine learning that considers thousands of data points for individual properties, local markets, and individual seller preferences. Users are asked a few simple questions about their home and their top objectives during the sales process. Proprietary technology then evaluates the top local real estate agents, institutional buyers, discount agent models, and more to match the seller with the option that will best help them achieve their goals. The service also provides a personal concierge to assist the homeowner through their selling experience.

Provided Services:
Back-End development, DevOps setup

• Development of a quiz, calculating the approximate cost of a house, compiling a list of the 5 best realtors, displaying the nearest recently sold houses on a map and statistics
• Implementation of a dashboard for sales department with agent statistics
• Implementation of the client's idea of selling 2 places from the list to agents
• Referral system development
• Server setup and optimization

Value:

• Received Inman Innovator Awards in 2019
• After we implemented the idea of ​​selling slots in Zip - in the first month, it delivered company $50k.
• We are currently working to reduce costs on AWS: In September, they paid $8400. Forecast for November - $7000. And we also plan to reduce it to $6600 this week
• Implemented a tool for salespeople to help identify agents. The service "guaranteed display" will be useful to them.

Challenges:

1. It was necessary to use MongoDB with setup backups, scaling and managing
2. The platform had limited calls to Salesforce per minute, so there was a need to increase it
3. The client wished to be able to view analytic reports
4. Needed to reach stable server working in a period of high customers activity (Tom Ferry summit etc.)

Solutions:

1. We decided to choose MongoDB Atlas cloud database service, which covers the mentioned demands
2. We chose Amazon SQS for the purpose to stretch in time all calls to Salesforce from all microservices.
3. Used Stitch to setup ingesting data every 30 minutes to Redshift database from MongoDB, Salesforce and PostgreSQL. Then we used Looker to create charts from Redshift tables.
4. Made a decision to set-up autoscaling instances in case if CPU utilization is over 80% during 3 minutes

Technologies & Tools: 

Languages: Java, JavaScript, Bash, SQL
DataBases: PostgreSql, MySql, MongoDb, Redshift, BigQuery
Tools and frameworks: Spring boot, Spring MVC, Spring JPA, Spring Security, NodeJs, React, Typescript, HTML, CSS, Docker, Jenkins, EC2, Amazon S3, Amazon RDS, Amazon Sqs, REST, Maven, Git, Nexus, GraphQl, Swagger, Stripe (Blackthorn), Salesforce, Apex, Graylog