A smart online math tutor serving thousands of users
Plario Adaptive Learning System
A smart online math tutor serving thousands of users
Plario Adaptive Learning System
A smart online math tutor serving thousands of users
The Client
Plario, Adaptive Learning System
The Client
Plario, Adaptive Learning System
Challenge
Our goal was to develop a smart adaptive learning system in a six months period that would meet the following requirements:
Challenge
Our goal was to develop a smart adaptive learning system in a six months period that would meet the following requirements:
Challenge
Our goal was to develop a smart adaptive learning system in a six months period that would meet the following requirements:
Stable and maintainable
Easy to scale
Easy to extend with new services
Smooth continuous delivery process
Logging, monitoring and alerts features
Expected number of users: 10,000
After the explosion of online learning services under conditions of social distancing, the system was scaled to accommodate up to 100,000 users just in 30 days
After the explosion of online learning services under conditions of social distancing, the system was scaled to accommodate up to 100,000 users just in 30 days
After the explosion of online learning services under conditions of social distancing, the system was scaled to accommodate up to 100,000 users just in 30 days
Approach
Approach
Approach
Build architecture
Set up orchestration tools
Run load tests
Fine-tune
Build architecture
Set up orchestration tools
Run load tests
Fine-tune
Build architecture
Set up orchestration tools
Run load tests
Fine-tune
To achieve the goals, the following technology stack was chosen:
■ Docker as the microservices containerizing solution ■ Kubernetes orchestration platform ■ Rancher 2.0 orchestration management tool ■ GitLab CI/C ■ JMeter as a load testing tool ■ ElasticSearch/Kibana for logging ■ Prometheus/Graphana for monitoring
To achieve the goals, the following technology stack was chosen:
■ Docker as the microservices containerizing solution ■ Kubernetes orchestration platform ■ Rancher 2.0 orchestration management tool ■ GitLab CI/C ■ JMeter as a load testing tool ■ ElasticSearch/Kibana for logging ■ Prometheus/Graphana for monitoring
To achieve the goals, the following technology stack was chosen:
■ Docker as the microservices containerizing solution ■ Kubernetes orchestration platform ■ Rancher 2.0 orchestration management tool ■ GitLab CI/C ■ JMeter as a load testing tool ■ ElasticSearch/Kibana for logging ■ Prometheus/Graphana for monitoring
Тhe Product
Plario.ru is a matching online course that helps high school students and college freshmen bridge their gaps in basic math skills.
Тhe Product
Plario.ru is a matching online course that helps high school students and college freshmen bridge their gaps in basic math skills.
Тhe Product
Plario.ru is a matching online course that helps high school students and college freshmen bridge their gaps in basic math skills.
Simple diagnostics of initial mastery level
Personalized learning trajectories based on the current skills
Real-time progress tracking
Algebraic expressions conversion Trigonometry Logarithm Rational equations and inequations Irrational equations and inequations Functions
Features
Rational equations and inequations
Irrational equations and inequations
Functions
The platform is an EdTech startup that involves active participation of the Tomsk State University scholars. The educational content was created by skilled mathematicians with the help of special content graining technologies by ENBISYS.
The platform is an EdTech startup that involves active participation of the Tomsk State University scholars. The educational content was created by skilled mathematicians with the help of special content graining technologies by ENBISYS.
The platform is an EdTech startup that involves active participation of the Tomsk State University scholars. The educational content was created by skilled mathematicians with the help of special content graining technologies by ENBISYS.
The assessment and adaptive learning algorithms are based on an ontology of interconnected basic math skills, some graphs comprising 54 units. The student's math level is tested in just 24 diagnostic problems.
The assessment and adaptive learning algorithms are based on an ontology of interconnected basic math skills, some graphs comprising 54 units. The student's math level is tested in just 24 diagnostic problems.
The assessment and adaptive learning algorithms are based on an ontology of interconnected basic math skills, some graphs comprising 54 units. The student's math level is tested in just 24 diagnostic problems.
Adaptivity
Having to deal with an enormous amount of analytics and algorithms, we applied Machine Learning and genetic algorithms, Bayesian Knowledge Tracing (BKT), hidden Markov processes.
BKT allows real-time assessment of the mastery level and creation of a dynamic personalized learning track.
Test results among university freshmen show dramatic increase in performance (see the chart).
Adaptivity
Having to deal with an enormous amount of analytics and algorithms, we applied Machine Learning and genetic algorithms, Bayesian Knowledge Tracing (BKT), hidden Markov processes.
BKT allows real-time assessment of the mastery level and creation of a dynamic personalized learning track.
Test results among university freshmen show dramatic increase in performance (see the chart).
Adaptivity
Having to deal with an enormous amount of analytics and algorithms, we applied Machine Learning and genetic algorithms, Bayesian Knowledge Tracing (BKT), hidden Markov processes.
BKT allows real-time assessment of the mastery level and creation of a dynamic personalized learning track.
Test results among university freshmen show dramatic increase in performance (see the chart).