Writing
RSSThoughts on software engineering, technology, and building digital products.
- Blog
What vLLM actually does and why your GPU is wasting memory without it
An open-source inference engine that enables fast, memory-efficient serving of Large Language Models at scale
- Blog
Building ReflectRAG - What Tutorials Won’t Tell You About RAG
A self-correcting RAG system from scratch. No LangChain. No LangGraph. Just raw Python, FAISS, and the OpenAI SDK.
- Blog
Thundering Herd
A concurrency bug in my Notion Image CDN caused a “thundering herd” cache stampede that could OOM-kill the service under real traffic—fixed by implementing a Go-style singleflight request deduplication.
- Blog
Vector Databases: Beyond Keyword Matching
Traditional databases rely on exact keyword matches. Vector Databases understand context and intent. This blog explains the architecture that makes modern AI retrieval possible.
- Blog
How AI Coding Tools Are Rewiring Software Development
Software development is changing fast. Agents write code, engineers review and orchestrate. Here's what agentic engineering looks like from the inside; the tools, the costs, and the new skills that matter.