A brief of how we built translate-transcription service for Skillshare.

Skillshare is an American online learning community for people who want to learn from educational videos. The courses, which are not accredited, are available through subscription. The majority of courses focus on interaction rather than lecturing, with the primary goal of learning by completing a project.

Skillshare needs to go beyond English spoken users but videos and transcriptions are in English. They want to extend their target, they need online translations for their video transcriptions, and this process needs to be automatic without affecting user experience.

What we did was to bring an automatic service to translate the transcription of a video the user is viewing.

Our solution brings what they need right now In this MVP you can see how a video is shown with a transcription in English, then the user can select his or her native language and easily can follow the lesson by reading translated transcription online.

The first time a user selects a language for a specific video, translation occurs and is stored for future users. This way the resources used are minimum and the process is automatic, without the need of somebody who runs a translation process each time a new video arrives.

This is our team, Zombieland CLI: Carlos in frontend and design, Ivan in frontend and testing, and myself, Leonardo in the backend programming in Python.

The architecture for this PROJECT uses AWS services specially A.P.I. Gateway and Lambda function using Python. The A.P.I was an A.P.I Rest, we used a get method to make the request, and then the A.P.I calls for the lambda function. The lambda function we created detects the video language to see if the translation requested is already stored and if not, calls a translate_text method.

We decided to work in this project for several reasons, the main of this was the challenge to use a cloud service like AWS and to create a real solution for a company who was needing this. We learned to learn at Holberton School so we like projects with new technologies because it is the perfect opportunity to grow as professionals.

Our project was scrum-based, we had weekly sprints, a daily meeting, and Sprint review on Fridays. In addition, we had Mandatory rules for the team like daily meetings with the Client representative, two daily team meetings to share individual research and to do pair programming to integrate codes. A special rule to avoid migration to production code without the approval of all the team. Finally, our timeline was of six weeks.

Our main challenge was architecture definition. This was because we had several options, and to select the optimal one we had to investigate technical details about how Skillshare is working right now. Initial architecture definition has struggled because of technical details not well known by the client representative. However, his trust in our decisions helped.

There was a lot of learning lessons for the team:

• To define architecture with an extended team
• Scrum helps to organize, prioritize, and develop. 
• Well defined team rules help during the journey.
• To use the right technology rather than use “the best”.

However, we want to say for sure than now we feel we are capable to learn what we need.

Leonardo Calderon, a Software developer with experience as a project manager to a wide range of businesses onsite and remotely. My primary interest is in high-level language applications with a focus on future development in data science and machine learning. My goals are to be a business-oriented software developer, an important part of key projects for a company and to use my skills as a specialist, consultant, or teacher to improve teams, processes, and business.

GitHub: https://github.com/leocjj/transcriptions-translations-Skillshare

MVP: https://transcriptions-translations.s3.amazonaws.com/index.html

Landing page: https://transcriptions-translations.s3.amazonaws.com/landing/index.html 

LinkedIn: https://linkedin.com/in/leonardocj