About Me

Machine Learning Engineer | GenAI & TTS Specialist

Hi, I’m Matteo – a Machine Learning Engineer passionate about bridging the gap between human interaction and AI. My expertise lies in Speech Synthesis (TTS) and Natural Language Processing (NLP), particularly in building Retrieval-Augmented Generation (RAG) systems that are both innovative and robust.

Currently, I focus on deploying large-scale Generative AI solutions in enterprise environments, ensuring they are not just "smart" but also safe, scalable, and compliant. I thrive on solving complex challenges—whether it's giving a voice to a machine or making financial knowledge accessible through conversation.

Experience

Open to interesting collaborations in AI! Feel free to reach out via email.

Professional Timeline

Senior Machine Learning Engineer @ E.Sun Financial Holding
Mar 2023 - Present | Taipei, Taiwan
  • GenAI Investment Assistant (投資 i-chat): Led the development of Taiwan's first conversational AI for finance (launched July 2025). Built a RAG-based engine that combines internal research with real-time market data.
  • Safety & Compliance: Designed a multi-layered safety architecture (Model Armor, role-based filters) to prevent hallucinations and ensure compliance.
  • Hybrid Cloud: Migrated services to a scalable GCP architecture (Cloud Run, AlloyDB) to handle high-concurrency requests.
Machine Learning Engineer @ E.Sun Financial Holding
July 2021 - Mar 2023 | Taipei, Taiwan
  • Text-to-Speech (TTS): Built an in-house TTS service (Tacotron2, VITS) from scratch with a custom 20-hour voice dataset.
  • Chatbot & NLP: Managed a cluster of 12+ BERT-based internal chatbots and deployed high-throughput API gateways.
  • Recommender System: Developed a Spark-based DeepFM recommendation engine serving 9 million users.
Software Engineer Intern @ Walsin Lihwa Corp.
Apr 2018 - July 2018 | Taipei, Taiwan
  • Implemented Text Summarization using PageRank and Bi-LSTM for industrial factory reports.
  • Automated talent data extraction pipelines for HR analytics.
M.S. in Computer Science
National Tsing-Hua University | Hsinchu, Taiwan
B.A. in Computer Science & Information Management
Soochow University | Taipei, Taiwan

Skills

Generative AI & NLP
  • RAG Systems
  • LLM Fine-tuning (PEFT/LoRA)
  • Vector Databases
  • AI Safety & Guardrails
Machine Learning & Speech
  • PyTorch
  • Text-to-Speech (VITS/Tacotron)
  • Recommender Systems
  • Model Optimization (ONNX)
Cloud & MLOps
  • Google Cloud Platform (GCP)
  • Docker & Kubernetes
  • CI/CD Pipelines
Languages
  • Python
  • SQL
  • C++
  • Shell Scripting

Projects

TTS System

Text to Speech System

Built an in-house TTS service with 0.3 RTF using Tacotron and VITS architectures. Curated a 20-hour voice dataset for training high-fidelity models.

Python PyTorch VITS Tacotron
Paper
Singing Synthesis

Singing Voice Synthesis

A two-part harmony singing synthesis system that generates vocal harmonies from a main melody. Won 3rd place in CTCI AI Arts Competition.

WORLD Vocoder DSP MusicXML
Poster Site
Lawyer Rec Sys

Lawyer Recommendation

Analyzed 9 million court cases to build a recommendation system matching users with suitable lawyers. Visualized data using Tableau.

Python Pandas Tableau
Demo Site
AI Cloze Test

AI Cloze Test System

Achieved 80% accuracy in automated cloze tests using BERT and Word2Vec. Ranked 1st in NTHU AI competition.

BERT Transformers Word2Vec
Code
StrikeZone AR

StrikeZone AR

An AR-based virtual strike zone application for baseball training. Built with Unity3D and Vuforia. Won 2nd place in ICT Awards.

Unity3D C# Vuforia
Demo News

Interests

Outside of engineering, I am a music lover, producer, and sound enthusiast. I enjoy exploring R&B and classical genres, and I mix my own covers.

Latest Music Work

【好不容易 Finally】告五人 - Cover

Produced and mixed in Logic Pro. Exploring R&B and pop arrangements.

Watch on YouTube 🔗

Resume

View my full resume or download a PDF copy.