50ms
avg response on the core ordering path
Personal Homepage / Journal
Stev5
Personal homepage, journal, and selected systems work on backend architecture, practical AI engineering, and the quiet discipline of reliable delivery.
92%
top-3 retrieval hit rate after retrieval tuning
85%+
coverage on core modules with AI-assisted testing
Java
Spring Boot
MySQL
Redis
RocketMQ
LangChain
Docker
GitHub
Selected Work
Two systems that improved reliability, throughput, and engineering speed.
A focused selection of work that best represents how I approach architecture, performance, and delivery in production environments.
Knowledge Infrastructure
AI team knowledge platform built for production use
Designed and delivered a Spring Boot based knowledge system that turned fragmented internal material into an AI-assisted workflow teams could actually rely on.
Spring Boot
LangChain
Java
- Reworked retrieval with query rewriting, semantic chunking, and hybrid search.
- Introduced AI-assisted review and test generation into the delivery loop.
92%
top-3 retrieval hit rate after system tuning
High-Concurrency Commerce
Flash-sale ordering flow engineered for stable peak traffic
Built an order pipeline that used traffic shaping, consistency controls, and multi-level caching to keep the user path responsive under burst load.
MySQL
Redis
RocketMQ
- Used Redis and Lua to protect inventory and enforce one-order-per-user rules.
- Smoothed burst traffic with asynchronous processing and consistency recovery.
50ms
average response after queue-based peak shaving
Journal
A developing archive of field notes and system essays.
Ongoing writing and field notes on backend architecture, practical AI integration, and the discipline required to keep production systems calm.
Designing systems that stay quiet under pressure
Why queue shaping, cache boundaries, and operational simplicity usually matter more than visibly complicated architecture.
Archive in progress
What I keep, and discard, from AI engineering workflows
A working view on retrieval, automated review, and test generation when the goal is better delivery instead of better demos.
Archive in progress
Latency, load, and the craft of making fast feel stable
Notes from building high-concurrency paths where response time, consistency, and operational calm all have to coexist.
Archive in progress
Capabilities
Engineering that favors clarity, resilience, and real delivery.
01
Backend Architecture
Designing service boundaries, API flows, and delivery paths that stay understandable as systems grow.
02
Performance & Concurrency
Optimizing critical request paths, handling burst traffic, and controlling failure modes under pressure.
03
AI-Enabled Engineering
Applying retrieval, automated review, and test generation where they improve real engineering throughput.
Background
Systems thinking shaped by engineering discipline and competitive problem solving.
Currently focused on distributed systems, high-concurrency services, and practical AI integration for teams that care about stable execution.
National Scholarship Recipient
ICPC Silver Medalist
Resume Available on Request
Operating Priorities
Reliable delivery
Distributed systems, high-concurrency paths, and production behavior that stay understandable under load.
Practical AI integration
Retrieval, review, and testing workflows that improve engineering throughput without adding operational noise.
Public proof
Measured outcomes, GitHub history, recognition, and resume access available on request.
Recognition
- National Scholarship Recipient
- Enterprise Scholarship Recipient
- ICPC Silver Medalist
- National Second Prize, Robocom Developer Competition
- Outstanding Graduate
Contact
Building backend infrastructure, high-concurrency services, or practical AI systems?
Let's talk about work that values reliability, performance, and disciplined execution.