Alcune informazioni sono riportate in lingua inglese.
Chi sono
• Integrated LLM into content pipelines for multilingual processing, summarization, content filtering to prevent users from adult content.
• Reduced amount of content in data pipeline by implementing scheduling parsing of posts from RSS feed.
• Experimented with LLM models to choose the best cost-effective solution. Reduced expenses by 4.1x with refactoring.
• Fixed language misalignment and added multi-categorical response.
• Implemented an LLM-TTS system to automate answering questions in TikTokLive comments when the streamer leaves the stream.... Continua a leggere