Platform modernization
Migrate legacy stacks to cloud-native architectures that stay fast, observable, and cost-aware.
Cut Flexport infrastructure costs by 25% while improving p99 latency from 20s to 500ms.
Software architect & fractional CTO for AI teams.
Greg Konush partners with founders to turn prototypes into durable products, combining systems thinking, pragmatic AI, and hands-on engineering leadership.
Previously led platform modernization and marketplace launches at Flexport, delivered zero-trust security products at Illumio, and now ships resilient cloud platforms that keep AI features measurable and compliant.
- Currently
- Partnering with founders as a fractional CTO
- Based in
- San Francisco Bay Area
- Focus
- Resilient platforms, AI-assisted workflows, product velocity
How I help teams
From founder-led startups to enterprise platforms, I translate complex requirements into fast, observable, and secure systems. These are the lanes where I usually drive outcomes.
AI-enabled workflows
Design secure data flows and AI integrations that keep humans in the loop and operations compliant.
Built ML-powered automation for fraud and risk teams, saving thousands of support hours annually.
Founder partnerships
Translate product strategy into shipping roadmaps, leading small teams from prototype to durable release.
Helped Truewind close Top 50 accounting firms and raise Series A by delivering critical product modules.
Recent impact
Selected highlights from the last few years that show how I approach product velocity, platform rigor, and AI adoption.
Multi-market logistics launches
Led integration of Deliverr at Flexport, shipping multimodal freight and FTL marketplace offerings that expanded coverage.
Security and reliability at scale
Built vulnerability analysis tooling at Okta and zero-trust segmentation products at Illumio to harden enterprise stacks.
Data-first product thinking
Pair hands-on engineering with product feedback loops to keep AI features measurable, governable, and user-friendly.
Temporal Bun SDK hits 0.3.0
Announcing the @proompteng/temporal-bun-sdk 0.3.0 release with Effect powered workflows and Bun native examples.
Building a native Bun runtime for Temporal with Zig
How a Zig-powered native Bun addon brings Temporal workflows directly into the lab experiments ahead of the npm release.
Vercel AI SDK with Next.js
Shipping real-time AI flows in Next.js while keeping latency low, UI snappy, and observability first-class.
Kubernetes Installation in Harvester
Standing up controlled Kubernetes clusters on Harvester with GitOps guardrails and day-two automation baked in.
Astro for personal website
I tried to use astro my own website, great framework all around
Rails with GraphQL
Combining the Rails toolkit with GraphQL schemas to speed up internal tooling, and how to keep it maintainable.
Monorepo setup with NX CLI tooling
Attempted to use NX CLI for my own needs but my monorepo is multi-language
Modern Tailwind CSS and Next.js setup
Updated version of tailwindcss v4 setup in nextjs, outdated, you can just create next with tailwindcss
Work
- Aug 2024 - Present
Truewind
Founding Software Engineer
- Jul 2021 - Jul 2024
Flexport
Staff Software Engineer
- Jul 2019 - Jul 2021
Illumio
Senior Software Engineer
- Aug 2018 - Jul 2019
Okta
Senior Software Engineer
- Sep 2017 - Aug 2018
Eventbrite
Software Engineer
- Mar 2017 - Sep 2017
Ticketfly
Software Engineer
- Sep 2015 - Mar 2017
Narvar
Software Development Engineer in Test
What I’m building
Engagements and experiments that keep my curiosity sharp.
proompteng
Kubernetes control plane for AI agents
Learn moreExperimentation Lab
Experiments with different tooling across stack, do check it out
Learn moreMilkshake
CLI to generate images with Nano Banana that gets copied into clipboard
Learn moreMLX Dev Agent
Local first agent that utilizes Apples M chips
Learn moreTools I rely on
The stack that keeps the work calm, fast, and resilient.
Editor
Neovim with AstroNvim. Codex and Claude Code as agents
Stack
TypeScript, Go, Kubernetes, Python, NextJS, PostgreSQL
Workflow
Shape Up work ahead, do deep focused sessions after