Glossika Builds Language-Learning Platform on AWS to Serve Users in 148 Countries

Glossika started out producing language learning books and transitioned to an online model in 2017, followed by a mobile app in 2022. Upon going digital, Glossika chose Amazon Web Services (AWS) to build its IT infrastructure. Sheena Chen, chief operations officer at Glossika, says, “As a company striving for worldwide product adoption, we need cloud technology that can scale with us. AWS makes it easy to purchase the infrastructure we need right now and to adjust as we expand. AWS is also feature-rich and highly configurable, with an intuitive user console—all of which facilitates our growth as a startup in a sustainable way.”

Glossika currently serves customers in 148 countries who are learning over 60 different languages. Its courses guide users through a massive database of sentences in their target language(s). Sentences gradually increase in difficulty and are accompanied by recordings from native speakers. Users can make adjustments to each course in accordance with their preferences. For instance, someone interested in learning Japanese to read manga comics could choose to not practice sentences about working in an office.  

Glossika’s application works by organizing sentences according to the type of grammar/syntax they contain and how well a user is retaining them. At present, Glossika has uploaded about 6,000 unique sentences per language for learners to practice, and the company expects that figure to double in the next 2–3 years. Glossika uses Amazon Relational Database Service (Amazon RDS) with autoscaling enabled to store more than 350,000 sentences and their translations, Amazon Simple Storage Service (Amazon S3) to store over 25 million user-uploaded audio recordings of sentences cost-effectively, and Amazon ElastiCache for Redis as a low-latency caching service. Its adaptive learning algorithms run on Amazon Elastic Compute Cloud (Amazon EC2) instances.

Campbell elaborates, “Our algorithm considers several factors such as how recently a student learned certain information and how well they’re retaining it. If they just learned a structure yesterday or are struggling to replicate the sentence independently, students will see that structure more frequently.”  

To serve its global customers, Glossika relies on Amazon CloudFront as a content delivery network. The company is headquartered in Taiwan, but most of its customers live in the United States or Europe. To manage incoming traffic from around the world, its engineers built high-availability architecture using Elastic Load Balancing.

Campbell says, “The stability of AWS services is excellent. With active paying users in 148 countries, this is especially important for us. With AWS, we’re confident that our users have a reliable experience on our app no matter where they’re located.”

Glossika’s users continue to offer praise for the application as well. One learner comments, “Nothing else combines comprehensible natural audio input and active language production with nearly the same amount of practice as Glossika; every session is challenging, but always comprehensible and rewarding.”

In addition to scaling its business to add more users in more countries, Glossika recently launched a beta version of its Viva project to expand Glossika’s content offerings and preserve endangered and/or “minorized” languages such as Gaelic and Hakka.

While Glossika works with professional translators and voice actors to produce content, Viva crowdsources this information from users around the world who record and document their native languages in Glossika’s database. Participants who upload recordings of their language not only help preserve lesser-known languages and dialects, but also earn a share of the subscription revenue generated from learners studying that language.  

With an eye to expansion, Glossika is constantly innovating and developing new features that improve the learning process. One potential future feature would utilize machine-learning models in Amazon SageMaker to analyze audio files hosted on Glossika. This analysis would generate two colored lines above the text of a given sentence: one color showing the intonation of the native speaker’s recording and another showing the intonation of a user's uploaded recording. This information would let users see where their rhythm and intonation diverge from the native speaker’s. Learners can use the analysis to independently assess their speaking ability in any target language, helping them to improve their pronunciation and allowing them to more objectively assess how natural they sound.

Glossika has big plans for its global business and will continue to rely on AWS as its cloud provider in the next phase of its journey. Chen concludes, “AWS is reliable and easy to use. Because AWS has efficiently taken care of server- and security-related issues, our engineers have been able to focus completely on product development since day one. We look forward to growing our business further on AWS.”