About SteerScale
Production engineering, without the layers.
SteerScale is a senior-engineer-led consultancy founded by Barnabás Szőke. The person assessing the system is the person doing the engineering.
The practice stays small on purpose: one or two engagements at a time, written status every week, and no handoff from the person you spoke to first to someone you did not.
- EXPERIENCE
- 15+ years, production systems
- BASED
- Budapest · remote across Europe
- DOMAINS
- Streaming, retail, fintech, aviation, gaming, SaaS
- FOCUS
- AI systems · distributed systems · cloud reliability
NO SYNTHETIC PORTRAIT · ADD AUTHENTIC PHOTOGRAPHY IF AVAILABLE
01 / Philosophy
Four working principles.
- 01
Evidence before recommendations.
Measure first. Traces, cost data and controlled comparisons come before opinions about architecture.
- 02
Production behaviour beats diagrams.
Architecture must survive reality: load, partial failure, retries, rate limits and the cost of every call.
- 03
Complexity must earn its cost.
Prefer simpler systems where possible. Most production failures have boring causes; solve those first.
- 04
Ownership remains with the customer.
Engineering should leave the customer's team stronger: written decisions, code in their repo, no dependency on us.
02 / Selected production work
Where the production hours come from.
Backend, infrastructure and reactive-systems work across video streaming, retail, fintech, gaming, aviation and SaaS.
- 2024 — nowSteerScaleFounder & Senior Software Architect · AI systems, distributed systems, cloud reliability
- 2022 — nowSky / NBCUniversalGlobal video streaming platform (Peacock, NOW, Sky Showtime) · high-throughput backend services
- 2019 — 2022TescoTechnical lead · price transformation and payment platforms
- 2018 — 2019Engage Technology PartnersTech lead · vendor and agency back-office SaaS
- 2017 — 2018Ladbrokes CoralReal-time CRM big-data platform
- 2016 — 2017ClearScoreCredit score service · reactive microservices, Java / Scala
- 2015 — 2016UnderwriteMeInsurance comparison platform
- 2014 — 2015GamesysRemote game platform · Virgin Poker on AWS
- 2013 — 2014Lufthansa SystemsNetLine/Ops++ airline operations control
- 2010 — 2013evosoft (Siemens)Rail-control diagnostic systems
03 / Current focus
Built on 15+ years of production engineering.
The core of the practice remains backend, distributed-systems and cloud engineering. A growing share of current work applies that foundation to AI systems: LLM orchestration, model selection, inference economics, evaluation, reliability and observability. For teams already running an LLM or agent workflow, the AI Systems Audit offers a fixed-scope way to start.