Agent platform architecture
Designing core platform features for ADP, including workflow orchestration, model invocation scheduling, and quota-safe enterprise execution paths.
Building enterprise AI products from model capability to production reliability.
I am Krislliu, an AI backend engineer currently working on Tencent Cloud ADP, CodeBuddy, and WorkBuddy. I focus on LLM infrastructure, prompt systems, usage billing, and resilient API services that survive real customer traffic.
What I am working on now
Usage and billing integrity
Building usage metering, cost attribution, and enterprise-facing usage query flows for CodeBuddy and WorkBuddy open platform products.
Prompt and response quality loops
Shipping prompt optimization capabilities that improve generation consistency while preserving throughput under production constraints.
Projects with concrete shipping scope
Tencent Cloud Agent Development Platform
Led independent site development, built blog and ad system modules, and fixed production SEO issues including sitemap and robot directives. Also shipped rate limiting, model dispatch, and billing capabilities for enterprise stability.
Open platform usage systems
Implemented usage billing and quota management, resolved pagination issues affecting key enterprise clients, and delivered real-time API services with around 200ms response behavior for engineering knowledge workflows.
Real-time translation with MOE + wait-k
Trained multilingual translation models with Fairseq, integrated ASR-facing interfaces, and improved low-resource Cantonese data coverage through augmentation and terminology-aware generation.
Career timeline
Tencent Cloud · AI Backend Engineer
Backend development for AI Development Platform and intelligent assistant products, with Python/Go microservices running on Docker and Kubernetes.
The Hong Kong Polytechnic University · Research Associate
Researched multilingual translation systems and contributed to assistive smart-glasses projects for visually impaired users.
Huawei · Software Engineer
Developed cloud-container logging and alerting flows with Go, Fluentd, Elasticsearch, and Kubernetes deployment workflows.
Technical depth and delivery scope
AI engineering
End-to-end LLM application delivery from prompt design to production traffic control.
- Fine-tuning
- RAG
- Prompt Engineering
- Multi-Agent
- Workflow Orchestration
Backend systems
Service architecture built for stability, observability, and tenant-safe scale.
- Python
- Go
- Microservices
- Rate Limiting
- Usage Billing
Cloud-native operations
Operational ownership from deployment to production troubleshooting.
- Docker
- Kubernetes
- Redis
- Elasticsearch
- Incident Debugging
Cross-team execution
Partnering with product, operations, and enterprise clients to move priorities into shipped systems.
- Requirement Scoping
- Customer Support
- API Integration
- Bilingual Communication
- Technical Writing
Start from the context you need
Resume
Resume / CV snapshot available on request.
GitHub
github.com/apocalypseyun
liugz97@gmail.com
Telegram
@ApocalypseYun
Building AI products that balance model quality and production constraints.
If you are working on enterprise AI infrastructure, developer tooling, or multilingual systems, I am open to discussing product collaboration, platform engineering, and applied research.
Direct channels
Email: liugz97@gmail.com
Telegram: @ApocalypseYun
GitHub: github.com/apocalypseyun