Charles Yang isCharles

Hi, I'm Charles Yang

Java backend engineer, AI speech researcher, and BUPT graduate student based in Beijing.

I care about reliable backend systems, practical AI tooling, and the engineering details that turn demos into something people can actually use.

Java backend systems: Spring Boot services, REST APIs, authentication, caching, and deployment.
High-concurrency patterns: Redis cache strategies, atomic Lua scripts, async queues, and traffic smoothing.
Data and infrastructure: MySQL, Redis, RocketMQ, PostgreSQL, Nginx, Docker, JWT, and Cloudflare deployment.
AI product engineering: LLM provider integration, resume analysis, RAG-style knowledge bases, and practical AI tooling.

mode: understand the system -> find the bottleneck -> make it reliable

_{Shibuya Sky, Tokyo}

Project	What it is	Engineering Highlights
TimeGrocery	A city-local grocery and deals platform for shop discovery, voucher flash sales, and order workflows.	Spring Boot, Vue 3, Redis, RocketMQ, MySQL, multi-level caching, cache penetration/breakdown protection, asynchronous order processing, Lua-based atomic stock deduction, load testing, and observability-oriented performance work.
OfferMate	An AI interview preparation workspace adapted from Interview-Guide and deployed as a protected public demo.	Spring Boot, React, PostgreSQL, Redis, MinIO, LLM provider configuration, resume analysis, mock interviews, RAG-style knowledge base, Docker deployment, Cloudflare Tunnel, and protected public access. Demo

Work	Venue	Focus
StableTTS: Towards Efficient Denoising Acoustic Decoder for Text-to-Speech Synthesis with Consistency Flow Matching	2025 IEEE International Conference on Acoustics, Speech, and Signal Processing Workshops (ICASSPW)	Efficient text-to-speech acoustic decoding with consistency flow matching for faster, stable synthesis.
VoxAging: Continuously Tracking Speaker Aging with a Large-Scale Longitudinal Dataset in English and Mandarin	Interspeech, 2025	Large-scale longitudinal speaker aging study across English and Mandarin speech, spanning 293 speakers and up to 17 years of weekly recordings.

Languages

Backend

Data / Middleware / Infra

AI / Research

Less noise. More shipped experiments.

_{Open to collaboration on AI products, practical ML tooling, and clean web systems.}