Theo Yang

AI Engineer | Enterprise AI Applications · Data & Risk

Status

Approved U.S. immigrant visa. Open to relocation in the U.S.

Target Roles

AI Engineer · AI Application Engineer · Applied AI Engineer · Forward Deployed Engineer · AI Solutions Engineer

Professional Summary

AI engineer with a 16-year career spanning data engineering, machine learning and risk systems, enterprise data platforms, and AI applications. Hands-on with RAG pipelines, LoRA fine-tuning, multi-agent orchestration, and LLM evaluation in production, regulated environments. Proven pattern: validate an AI system end-to-end, abstract the capability for scale, and drive measurable business outcomes.

Technical Skills

LLM / AI

RAGLangChainReAct agentsLoRAInstruction fine-tuningPrompt engineeringKnowledge distillationMulti-agent workflows

Evaluation & Serving

Recall@kFaithfulnessHallucination controlHuman-in-the-loop reviewModel monitoringDrift detectionA/B testing

Data / ML

PythonSQLETLFeature engineeringXGBoostFAISSHadoopSpark MLlib

Domains

Fraud detectionCredit scoring & riskNL2SQL / ChatBIUser profilingLogistics analytics

Work Experience

China Telecom

Oct. 2023 – Present

Major telecommunications operator

AI Engineer — Fraud Detection Agent

  • Built a three-layer fraud-detection architecture combining rules, NLP/NER, and LLM+RAG for regulated environments.
  • Fine-tuned Qwen-32B with LoRA and solved data residency through one-way knowledge distillation and vector export.
  • Built a hybrid-retrieval fraud knowledge base and standardized API serving 15M calls/day at <200ms and 99.9% uptime.
  • Reduced case processing from 2–3 hours to 20 minutes, reached 85% officer adoption, and scaled to 25+ enterprise and device-manufacturer clients.

Meituan

Jan. 2019 – Oct. 2023

On-demand services & logistics platform

AI Engineer — ChatBI (NL2SQL) & Logistics Analytics

  • Built a ChatBI / NL2SQL system and replaced a pure-LLM path with three-tier routing, guided clarification, and a six-tool ReAct agent.
  • Improved query accuracy from 60% to 85% and reduced data-team workload by 90%.
  • Built a real-time logistics early-warning and diagnosis system that improved on-time rate, labor efficiency, sorting cost, and incident reduction.

Xiaomi Digital Technology

Oct. 2015 – Jan. 2019

Consumer electronics & IoT company

Data / AI Lead — Credit Scoring & Risk Platform

  • Built a thin-file credit-scoring platform with 1,500 behavioral and device features and an XGBoost model reaching AUC 0.75+.
  • Served 500M+ scoring calls and 2M+ users through APIs used by 40+ financial institutions.
  • Supported 20B+ RMB in lending and improved partner risk-approval rates by 10–20%.

JD.com

Jan. 2012 – Sep. 2015

E-commerce & logistics platform

Data Mining Lead — User Profiling, Recommendation & Risk

  • Led a five-person team building 360° user profiling and multi-channel recommendation capabilities.
  • Built recommendation and blacklist-mining systems for JD Baitiao consumer credit.
  • Built a big-data mining platform with Hadoop 2.0, Spark MLlib, and Mahout.

Huawei

Oct. 2008 – Dec. 2011

Telecommunications & enterprise technology company

Data Engineer — ETL, Metrics & BI

  • Built ETL pipelines, metric systems, and BI reporting for large-scale telecom data.
  • Performed network-protocol and user-behavior analysis and supported decision-support systems on telecom signaling data.

Education

Communication University of China · Master of Business Administration (MBA)

2021 – 2023

Hefei University of Technology · B.Sc. in Computer Science & Application

2004 – 2008

Want to see the engineering depth behind the resume?

The case studies show the architecture, trade-offs, production constraints, and measurable impact behind the headline metrics.

View Case StudiesContact Me