junluo yang

I’m an Algorithm Engineer who loves turning messy data into decisions, and then shipping reliable systems. My background blends logistics-related graduate training with hands-on work across data analytics, machine learning, and deep learning.

On the data side, I’m strong in Python’s data ecosystem (Pandas, NumPy, SciPy, SQL) and I build analysis that can survive real business constraints. I also work end-to-end with data cleaning, feature engineering, and A/B testing design, and I use Tableau to deliver interactive dashboards for stakeholders.

On the engineering side, I focus on performance and high concurrency: MySQL design and optimization, Redis caching and tuning, and message middleware like Kafka/RocketMQ. I develop APIs and high-performance async services using FastAPI, and I’m comfortable designing RESTful interfaces that integrate smoothly into production workflows.

For algorithms, I cover the full pipeline—from classic ML (decision tree, logistic regression, SVM, KNN, XGBoost, LightGBM) to deep learning (Transformer/LLM fine-tuning, prompt engineering, and Transformer-based NER with ModernBERT). At Chengdu Meierbei, I built an LLM + knowledge-base chat system with a configurable flow engine and a message-queue architecture, and I shipped high-precision entity recognition by replacing fragile pipelines with careful badcase-driven logits handling. I’ve also delivered multiple vision-generation and recognition projects using LoRA fine-tuning and DINO/Qwen-Image/FLUX.1 style models, and I enjoy using clear logic and reflection to iterate quickly toward measurable results.