• Collaborated with a 10-member team to integrate speech products (ASR, TTS) into Zalo and KiKi virtual assistant.• Developed text normalization modules for speech synthesis and recognition using hybrid methods (rule-based, language modeling, deep learning).• Improved ASR performance in Zalo ecosystem products (KiKi, LabanKey, ZingMP3) by optimizing the language model, boosting accuracy by 12%.• Optimized services deployed on Kubernetes (K8S), reducing error rates by 15%.• Collected, cleaned, and analyzed data, enhancing AI model accuracy for training tasks.

• Developed a semi-AI system to prevent spam comments on e-newspapers, utilizing Word2Vec for word similarity and classification models, significantly reducing manual labor.• Deployed AI services to servers using RESTful APIs, facilitating seamless integration with the CMS system.• Deployed AI services to servers using RESTful APIs, facilitating seamless integration with the CMS system. • Conducted experiments with Apache Hadoop and Spark for efficient data storage and computing, enhancing data processing capabilities.