BACKEND SYSTEMS & AI AGENTS
Charles Chi Le 乐驰
Backend & AI Agent Engineer
I build reliable cloud platforms and practical AI agents. My work spans Elasticsearch infrastructure, agent development, and backend systems at Alibaba Cloud, ByteDance, and Meituan.
A Duke University M.Eng. graduate and open-source contributor, I enjoy turning complex systems into useful, dependable software.
Zhejiang B.Sc.
Experience
Alibaba Cloud 阿里云
Java Backend & Agent Engineer
Feb–Nov 2026 Hangzhou, China
Computing Platform Division · AI Search Cloud Platform
- Elasticsearch PaaS & health events. Built control-plane capabilities across events, monitoring, inspections, and agents. Connected monitoring, log ingestion, and anomaly detection to generate about 1,500 health events per week across 400+ instances.
- ES Agent. Integrated authorization, tools, knowledge bases, and Skills for diagnostics, data insights, and instance management. Serves 150+ users and answers 2,000+ questions monthly.
- Intelligent inspections. Built Agent-first inspections with rule-based fallback, covering 16,000+ instances at approximately 99.9% success, with scheduled execution and structured reports.
- Skill evaluation & improvement. Built evaluation and iteration workflows across 12 fault scenarios, raising scores from about 40% to nearly 90%. Distilled 19 reusable Skill lessons from 541 cases, with human review before release.
ByteDance 字节跳动
Backend Engineering Intern · Go
May–Aug 2025 Hangzhou, China
Global E-commerce · Funds Center
- Built configurable billing workflows, reducing development time for standard requests from 4 to 0.5 person-days. Developed Elasticsearch and Hive dashboards to monitor more than 20 million orders daily.
- Used RocketMQ and EaseJob to retry failed settlement tasks, improving billing–settlement consistency within SLA from 80% to 99.99%+. Developed MySQL–Elasticsearch consistency checks.
Meituan 美团
Backend Engineering Intern · Java
May–Aug 2024 Beijing, China
Core Local Commerce · Search Advertising
- Developed search-ad serving logic for pharmaceutical brands, including CPT/CPM sales logic and merchant/brand billing.
- Optimized an inverted-index retrieval API from 500 to 2,000 QPS, reducing tail latency from 181.9 ms to 24.7 ms. Compared traffic allocation and significance reporting across A/B experiment platforms.
Education
Duke University
M.Eng. in Electrical and Computer Engineering
Software Development
Zhejiang University
B.Sc. in Biomedical Sciences
Bioinformatics
Open source
MatrixOne
RepositoryGo database development · Aug–Dec 2024
Developed an LLM-based secondary index supporting document chunking, embeddings, and vector-based RAG through dynamically created SQL tables.
Read pull request #18809ModelCache
RepositoryLLM backend components · Oct–Dec 2024
Contributed a reranker module, additional embedding models, and migration to the FastAPI framework.
Explore the projectTools I work with
Languages
Java · Go · Python · C/C++ · SQL
Backend & infrastructure
Elasticsearch · MySQL · Redis · Spring · Linux · Git · Kubernetes
AI & development
Agent development · RAG · PyTorch · GDB
Always learning, always building.
This notebook is where I keep my computer science course notes, programming practice, and engineering discoveries.