Writing Code in the Early Days
My career started with C and C++ — writing high-frequency transaction platforms for major financial institutions. This was the era of SET protocols for card payment security, when every microsecond mattered and every line of code had to be bulletproof. There was no Kubernetes, no Docker, no cloud. Just bare metal, tight loops, and the understanding that if your code failed, real money was at stake.
That experience taught me something I still carry: the best systems are the ones built with deep understanding of the problem domain, not just the technology.
The CCIE Years
Earning CCIE #4251 — one of the earliest Cisco Certified Internetwork Expert certifications ever issued — changed my career trajectory. It wasn't just a certification. It was an immersion in how networks actually work at the protocol level. BGP, OSPF, IS-IS, MPLS, spanning tree, QoS — understanding these protocols deeply meant understanding the infrastructure that everything else runs on.
That understanding became the foundation for everything that followed. When you've debugged BGP route leaks at 3 AM across a global network, you develop an instinct for how complex systems fail. That instinct is worth more than any framework or methodology.
Scaling to Billions
The middle years of my career were defined by scale. Leading the consolidation of 42 datacenters to 6. Migrating 125,000 network devices to software-defined fabric. Managing $180M annual infrastructure budgets. Delivering a $1B+ modernization program 40% ahead of schedule.
At that scale, individual technical decisions compound. A 1% improvement in network efficiency across 10,000 applications saves millions. A 5% reduction in incident response time across 200,000 architecture elements prevents outages that would cost hundreds of millions. The stakes were high, and the margin for error was razor-thin.
The AI Pivot
The pivot to AI wasn't sudden — it was gradual. First came the GenAI RAG platform that 5,000+ users adopted for daily queries. Then the autonomous SOC with agentic AI. Then the realization that every AI system I'd built or evaluated had the same limitation: they found correlations, not causes.
Networks don't fail because of correlations. Applications don't slow down because of statistical associations. Security threats don't emerge because of pattern matches. They happen because of causal mechanisms — and understanding those mechanisms requires a fundamentally different kind of AI.
Building netcausal.ai
In 2025, I founded netcausal.ai to close that gap. The thesis is simple: enterprise infrastructure deserves AI that understands why things happen, not just what happened. Causal inference, structural causal models, do-calculus — these are the mathematical tools that make that possible.
Today we're shipping 13 AI products — from managed SSE and CPaaS platforms to causal inference engines and autonomous network intelligence. Every one of them is built on the same principle that started with writing C code 27 years ago: understand the problem deeply, then build the system that solves it.
The path was never straight. But every step — from financial transaction code to CCIE to $1B programs to causal AI — built toward the same insight: the best technology is technology that truly understands the world it's operating in.