LATEST_POSTS
How I Cut MCP Token Usage by 50% Without Hurting Agent Quality
Building MCP integrations with my AI team taught me where context tokens actually go. Here are the schema, payload, caching, and tracing changes that cut our MCP token usage by about 50% without sacrificing reliability.
Why Retail Needs a Content Supply Chain, Not Just AI
A product goes viral and demand skyrockets within hours. Whether retailers benefit is decided in that window. AI alone doesn't solve this; a functioning content supply chain does.
Fixing Gmail Email Clipping: HTML Minification with AST Parsing and LLM Refactoring
Stop Gmail from truncating your emails at the 102 KB limit. Learn how to combine Tree-sitter Concrete Syntax Tree (CST) parsing, MSO client-profile pruning, and constrained JSON patch refactoring to achieve 20–35% byte savings without breaking complex email layouts.
ABOUT_ME
> I'm a Senior Technical Architect at Adobe building intelligent systems that power creative workflows at scale.
> I specialize in AI Agents , LLM Integration , and Architecture — turning complex AI capabilities into reliable production systems.
> When I'm not shipping code at Adobe, I write about software architecture, AI engineering patterns, and building things that last.
$ cat interests.txt
Sr. Technical Architect
Adobe
AI Systems & Architecture
Building in prod
TECH_STACK
Technologies I work with every day
