I build systems that survive exponential growth, handle ambiguity, and are resilient to human optimism.-- KG.
Staff / Principal Software Engineer • AI Systems • Distributed Architecture • Ruby • Python • TypeScript • PostgreSQL
Resume • Blog • Consulting • LinkedIn • GitHub • RubyGems • SlideShare
I’m a lifelong coder, with a diverse career, who is currently a hands-on Staff / Principal-level software engineer and former CTO who builds and delivers to production reliable, scalable, and fault-tolerant systems.
My background spans distributed systems, PostgreSQL, Ruby/Rails, infrastructure, developer productivity, and, more recently, applied AI systems: retrieval pipelines, eval harnesses, production automation, and embedding-based services.
I like hard problems with real constraints:
- scaling transactional systems
- eliminating operational risk
- improving developer experience
- shipping AI features without turning the product into a probabilistic mess
- Build and scale large distributed backend systems with strong reliability and observability baked in
- Design production AI workflows around real business use cases
- Improve CI/CD, onboarding, deployment safety, developer productivity, and operational simplicity
- Comfortable in all levels of abstractions, from back-of-the-napkin architecture, to nitty-gritty code details.
- Turn messy systems into maintainable ones, and have successfully rewritten systems from scratch under two months.
- Practice 90%+ test coverage and it will convince you to do the same.
Important
- 4x CTO / technical leader with decades of experience across startups and growth-stage companies
- Helped scale a consumer platform from a fragile prototype to a Rails system handling massive growth with 99.99% uptime
- Achieved 100% uptime during ACA open enrollment at HealthSherpa under major traffic spikes
- Averted a potentially catastrophic PostgreSQL transaction wraparound incident on a 27TB database
- Built applied AI systems in production, including:
- recommendation systems using Python, PostgreSQL, and pgvector
- embedding refresh pipelines with zero-downtime cutovers
- internal AI automation for workflow extraction and follow-through
- Created and maintained Ruby open source projects with 200M+ cumulative gem downloads
- Co-created early Ruby rules for Bazel, contributing to the path toward modern Ruby support in Bazel ecosystems
- Conference speaker on scalability, PostgreSQL, Ruby, and systems architecture
- Applied AI engineering
- LLM product architecture
- Evaluation-driven development
- Retrieval and ranking systems, in particular Power Law distribution
- PostgreSQL performance, scalability, replication architecture, and security
- Ruby, Python, TypeScript, C/C++, BASH, Rust
- Distributed systems, with high availability, fault tolerance, and auto-healing architectures
- Developer tooling, automated onboarding, and build systems
- Gaming Algorithms, Mini-Max, Alpha Beta Pruning
- Networking Systems, CISCO routers, VLANs, etc.
- Hardware, ESP32, Arduino, Raspberry PI, and more.
A few projects that reflect the range of work I enjoy (the top ones are most recent):
-
inquirex — A DSL for building lead qualification forms that gather non-sensitive data, forms that can branch, skip irrelevant questions, compute running totals (such as price of service), and integrate with an LLM to allow the user to describe their relevant situation in their own words, from which LLM extracts structured answers, and skips them later on. The gems are all open source, but they are the basis for the soon-to-be-launched SaaS service @ qualified.at.
-
gomoku-ansi-c
ANSI-C-based game engine for 5-in-a-row (or Gomoku) that runs in a TTY terminal, and you can play using arrow keys.Or, you can use the online version running on Google Cloud Run, which offers a ReactJS frontend talking JSON API to a FastAPI Python server with PostgreSQL connectivity, which proxies game moves to a cluster of stateless single-threaded C99/HTTPD backends behind Envoy reverse proxy. Game API includes the entire game state in the JSON hash. You can play the game here: app.gomoku.games
-
simple-feed
Redis-backed social feed architecture similar to Twitter, with O(1) feed read-time property. scales to tens of millions of users. -
colored2
Terminal color support for Ruby -
sym
Symmetric encryption via CLI and Ruby API -
puma-daemon
Puma daemonization support -
cmake-project-template
C++ / CMake starter project with sensible defaults -
bashmatic
Bashmatic — 900+ DSL functions written in BASH for readable script feedback, terminal graphics, headers, and more.
You can also browse the full collection of public gems written in Ruby here:
I write and speak about:
- PostgreSQL scaling
- Ruby and systems design
- developer productivity
- DevOps and release engineering
- applied AI systems
Some links:
Originally from Kharkiv, Ukraine. Now based in the San Francisco Bay Area.
I graduated Suma Cum Laude from Monash University, majoring in pure mathematics and applied statistics, which permanently damaged my tolerance for fuzzy thinking.
- Agentica Group — parent company of my consulting one






