About

The Engineer Behind the Labyrinth

I'm Aditya Shubham. I turn ambiguous problems into working systems, and I use AI as a core part of how I build, debug, research, and ship. Most of what you see here went from idea to production fast because I treat AI as a collaborator, not a novelty.

I'm drawn to agentic systems and automation, workflows modeled as agent loops. Shopping Coach is a grounded AI assistant that recommends real products from a store's catalogue without hallucinating. JobAgent pairs a Go rules engine with a Python AI orchestrator. SD18, a full e-commerce platform, was designed and built end-to-end with AI in the loop.

I work in TypeScript, Rust, Go, and Python, picking the right tool per problem and shipping. I've published open-source packages used in production, including a Razorpay payment provider for Medusa, and I also run Elevate Strategy, a digital agency for sports and lifestyle brands.

How I Work

Build With AI

AI is in my daily loop, to think, prototype, debug, and research. It's how I take a vague problem to a working system in days, not months, without losing the plot on architecture or quality.

Agentic Systems

I think in agent loops: rules engines, orchestrators, and tool use, grounded in real data so the system reasons about what's actually true. JobAgent and Mellow Nova are built on exactly this, deciding, acting, and adapting rather than just responding.

Ship to Production

Prototypes are cheap; shipped is the point. I architect, build the infrastructure, and put it in front of real users, including e-commerce platforms handling real money and published open-source packages.

Right Tool, Fast

TypeScript for web and mobile, Rust for performance-critical tools, Go for concurrent systems, Python for AI and automation. I pick per problem and move.

Background

Selected into a competitive engineering program (top 0.5% of 14,000 applicants). Built production systems in enterprise environments. Ran a digital agency delivering e-commerce and corporate web products.

Currently looking to build AI-native products, agentic systems, automation, and prototypes that make it to production, alongside people who ship as fast as they think.