Yiming (Polo) Huang

Partner with internal teams to turn ambiguous operational challenges into production-ready AI systems. Designed and built an autonomous code-repair agent that takes reported bugs from intake and RCA through isolated implementation, deterministic verification, independent review, and GitLab delivery, with resumable execution and policy-based safety gates. Own solutions end to end—from discovery and architecture to rollout, observability, and continuous improvement—while delivering agentic and data platforms across LangGraph, FastAPI, Airflow, and Kubernetes.

Designed a large-scale Similar Product Search platform serving 15M products using multi-modal embeddings (VGG + Sentence-BERT) with Faiss ANN indexing. Built an Image Copy Detection System (ResNet) achieving 95% accuracy. Engineered distributed crawling and task systems (Celery, RabbitMQ), reducing processing time by 90%.