These days, people are making their careers in content creation, especially on YouTube. So, people with large followings face a significant challenge in understanding their audience preferences and opinions. To build a proper content strategy, it's very important to understand your target audience's preferences, behaviors, and pain points. It's quite difficult for a high-profile content creator to extract meaningful insights from comments. The biggest thing is that YouTube revenue hinges on viewership and engagement.
So, we need to implement some solutions by leveraging the power of the machine learning system for analyzing feedback and identifying trends and patterns in viewers' behavior.
Presenting ๐๐๐ง๐ญ๐ข๐๐ฒ - ๐ ๐๐ก๐ซ๐จ๐ฆ๐ ๐๐ฅ๐ฎ๐ ๐ข๐ง, an e2e ml system to help the content creators to provide in-depth analysis of YouTube comments to enhance their content and engagement strategies.
Load the plugin in Google Chrome and browse YouTube comments. You will get small, colorful circles/highlighters ๐ด๐ต๐ข that represent the sentiment of comments.
๐ด - ๐๐๐ ๐๐ญ๐ข๐ฏ๐ ๐๐จ๐ฆ๐ฆ๐๐ง๐ญ
๐ต - ๐๐๐ฎ๐ญ๐ซ๐๐ฅ ๐๐จ๐ฆ๐ฆ๐๐ง๐ญ
๐ข - ๐๐จ๐ฌ๐ข๐ญ๐ข๐ฏ๐ ๐๐จ๐ฆ๐ฆ๐๐ง๐ญ
It also shows basic comment stats like total comments, total likes on comments, total replies, average likes, and average replies. Still working on introducing even more insightful features soon!
Frontend built using HTML, CSS, and JavaScript.
Backend developed with FastAPI and utilized caching to minimize latency.
โข Data versioning and data pipeline achieved with DVC
โข Used MLFlow for experiment tracking
โข DVC, MLFlow, and GitHub centralized with Dagshub
โข CI/CD workflow automated using GitHub Actions for end-to-end deployment.
โข The model deployed on AWS leveraged Application Load Balancer (ALB), Auto Scaling Group (ASG), ECR, S3, and CodeDeploy to ensure efficient and reliable model deployment.
Clone, tweak, and deployโitโs all yours to explore!
And don't forget to star โญ and fork ๐


https://drive.google.com/file/d/16kTXVFD3Rvm7zt2m0CEhsg3iYvC38914/view?usp=drive_link