Engineering intelligence at the hardware & system layer.
Deep architectural dissections of mixture-of-experts, local LLM serving, agentic protocols, and low-latency inference kernels.
Mohan
AI Systems Engineer & Full-Stack Architect
Chennai / Bangalore, India · UTC+5:30
I specialize in operationalizing frontier foundation models—designing high-throughput inference serving pipelines (vLLM), building sandboxed Model Context Protocol (MCP) agents, and deploying self-hosted AI applications with zero vendor lock-in.
Publication Stack & Principles
Zero client-side runtime bloat. Pre-rendered HTML served directly from edge memory with sub-10ms TTFB.
No invasive fingerprinting or tracking scripts. Powered by cookie-less Cloudflare Web Analytics.
Multi-stage Alpine Docker container running on Coolify with automatic rebuilds on git push.
Every article is exportable as raw Markdown at /blog/[slug].md and indexed at /llms.txt.
Subscribe via standard RSS or AI web crawlers.