Dmitry Nichiporchik
Chief Technology Officer, Yarvixo. LinkedIn
Accountable for what the numbers on this site claim: the multiplayer backends behind 52,000 and 39,000 peak concurrent players, the on-chain ownership layer in a live game, and the engineering writing here.
What he works on
Written as what the work is, rather than as a list of years.
Dmitry is CTO at Yarvixo, where he owns the technical decisions on the game and real-time side of the business: server-authoritative session services, matchmaking and rating systems, connection-driven autoscaling on Kubernetes, and the boundary between what a game keeps off-chain and what a player is meant to own on it.
The two figures this site leads with are his to answer for. A live mobile title reached 52,000 peak concurrent players at 62 ms average latency and 99.98% uptime; that client is not named. Metarun, which is named with the client's permission and installable from Google Play, peaked at 39,000 concurrent players during an event, at 60 ms average round-trip to the match service and 99.8% uptime in production, built by six engineers in four months.
He also built and runs ArendaOS, Yarvixo's own multi-tenant SaaS for rental fleets — which is the one system on this domain whose architecture a reader can inspect by opening the product, and the reason the writing here tends to argue from constraints in a database rather than from principles.
Where the depth is
The subjects he is the reviewer of record for, internally.
Authoritative multiplayer
Single-writer state ownership, reconnection without an advantage, anti-cheat as a structural requirement rather than a polish item.
Scaling connection-bound tiers
Why CPU is the wrong autoscaling signal, what headroom pod start time implies, and how to drain a pod holding a live match.
The on-chain boundary
Ownership on-chain, gameplay off it — and what it costs a project to put that line in the wrong place.
Multi-tenant SaaS
Tenancy as a schema decision, conflicts refused by the database rather than by application validation.
Retrieval and LLM integration
When retrieval is the answer and when fine-tuning is, and the evaluation work a demo does not need.
Client integration
The seam where multiplayer bugs live: prediction and reconciliation on the client against authority on the server.
What he has written here
Every article in the Insights section carries his byline.
Want the technical conversation rather than the sales one?
A 30-minute call about your concurrency target, your authority model and where your on-chain boundary should sit. No deck.
Message received
We’ll review your enquiry and respond within one business day.