About

Who I am

I’m Shashi Shankar — 15+ years in software and systems engineering: C++, embedded systems, the Android framework (AOSP and AAOS), Bluetooth and Wi-Fi platform engineering, and automotive production environments, with contributions to LLVM along the way. Most of my career has been spent where software meets hardware and a retry loop doesn’t fix failure.

What I’m building now

Production AI platforms, agent systems, evaluation infrastructure, and Edge AI deployment — the layer where models meet production systems and, increasingly, where they meet devices.

Why this combination matters

AI systems don’t fail at the model layer alone. They fail at interfaces, state, concurrency, memory, observability, thermal limits, lifecycle management, and operational control — exactly the places where embedded and platform engineering experience compounds. Very few people carry automotive-grade systems discipline into AI platform work. That’s the gap I’m standing in, on purpose.

The rules of this site

1. Every entry is an engineering artefact — code, a benchmark, a diagram, a design decision, a failure analysis, or an implementation note. They land as they’re made, often several in a day.
2. A substantial case study every week or two.
3. Every entry is labelled honestly: built, benchmarked, designed, investigated, or analysis. Commentary is never dressed up as implementation.
4. Numbers, diagrams, and trade-offs beat adjectives.

Evidence

• Projects — flagship builds, each shipping with architecture notes, code, and measurements as they land
• GitHub
• The log — the running record

Get in touch

• Email: shashi@shashforge.dev
• GitHub: github.com/shashforge

If you’re building serious AI products and need someone who thinks in systems, not demos — my inbox is open.