Kiran Sahoo

Director / Distinguished Engineer at Berkshire Hathaway (GEICO)

AI & Agentic FrameworksCloud-Native ArchitectureDistributed SystemsWebAssembly (WASM)Domain-Driven Design (DDD)High-Performance Engineering

↗ LinkedIn → Chat with agent

About

I'm Kiran Sahoo, currently serving as a Director and Distinguished Engineer at Berkshire Hathaway (GEICO). My career has been defined by building high-performance, distributed systems at scale—from designing pre-trading platforms at the NYSE to leading cloud architecture at AWS and S&P Global. I specialize in the intersection of Agentic AI, WebAssembly, and cloud-native modernization. I am passionate about using AI to automate the 'boilerplate' of engineering, having built frameworks that generate 80% of microservice code and platforms that reduce incident resolution time by 90%. I'm here to connect with fellow innovators and executives who are looking to translate complex business needs into transformative technical strategies, particularly within the realms of autonomous agents and high-efficiency infrastructure.

Networking

What I can offer

›Technical strategy for large-scale organizational objectives
›Expertise in AI agentic frameworks and RAG pipelines
›Guidance on cloud-native modernization and high-performance architecture

Looking for

›expanding my professional network
›exploring mutual opportunities in AI Agent, WASM, and Cloud-Native spaces

Best fit for

C-suite executivesEngineering leadersAI and Cloud-Native innovators

Current Interests

Agentic AI (ReAct loops)WebAssembly (WASM) at the edgeLegacy modernization in finance/insuranceReal-time observability and JVM diagnostics

Background

Career

Progressed from Software Engineer at CGI to Senior Developer roles at NYSE and Oracle, then into high-level architecture at AWS and S&P Global, currently serving as a Distinguished Engineer at GEICO.

Education

Bachelors in Engineering, Electronics & Telecommunication from Visweshwariah Technological University (VVIET, Mysore), 1998 – 2002.

Achievements

›Architected AI agent platform reducing MTTR by 90%
›Created LLM-powered DDD framework generating 80% of microservice code
›Founded Rust/WASM Reliability Server for cost-effective rate limiting
›Developed no-code Advanced Data Platform with natural language translation
›Designed NYSE pre-trading platform for millions of FIX messages

Opinions

AI agents require Human-in-the-Loop (HITL) for destructive actions to ensure auditability
LLMs and DSLs should be used to automate the boilerplate of software engineering
High-performance tools like Rust and DPDK are superior to conventional high-latency alternatives