// PRODUCTIZED · UE5 ARCHITECTURE AUDIT

Four weeks. Fixed fee. The unvarnished read on whether your multiplayer stack is going to ship.

Run by the team that spent five years building a UE5 MMORPG to 70 CCU per layer with Souls-tier combat fidelity — IRIS, Mover, GAS, custom dedicated servers, GPU-accelerated animation. The audit you'd run yourself if you'd already shipped it.

// 02
THE PROBLEM

Your internal lead knows there are problems. They don't have the calendar to call which ones are real.

You are six months, twelve months, eighteen months into a UE5 multiplayer build. Player count is climbing, the layer is starting to creak, IRIS is misbehaving on the systems you didn't migrate cleanly, GAS is bottlenecking somewhere nobody can find, the Mover API churn between engine minor versions has eaten two sprints. Your internal lead knows there are problems. They don't have the calendar, the distance from the codebase, or the comparison data points to call which problems are real, which are noise, and which will eat the next milestone.

An external read from a team that has been to the bottom of every one of these problems on a five-year UE5 MMO is the cheapest insurance you can buy before the next milestone gate.

A senior-led, four-week deep dive — delivered as a written audit you can show your CTO.

01

30–60 page audit document

PDF + editable source. Named risks, ranked. Evidence cited from your codebase.

02

Risk register with severity × likelihood × shipping-impact scoring

Spreadsheet, sortable.

03

Prioritized remediation plan

90-day, 180-day, 12-month horizons. Estimated effort per item.

04

Pod-sized engagement spec

If a follow-on engagement makes sense, it's scoped and priced before the audit ends. No surprise upsell.

05

Live readout

90-minute session with your engineering and production leadership, audit walked through end-to-end, questions answered.

// 04
SCOPE
MENU

Six audit surfaces. You pick three to five.

The audit is calibrated to four weeks of senior time. We don't try to audit everything; we audit the four to six surfaces that matter most for your build.

/01
Replication & netcode
  • IRIS adoption posture — what's migrated, what's still on Replication Graph, what should be either way
  • Replication frequency tuning, dormancy, relevancy at your CCU target
  • Authority model audit — server-auth vs. client-auth-with-rollback decision per system
  • Bandwidth profile — what's spending bytes that shouldn't be, what should be
  • Hit registration latency budget — the path from input to confirmed-on-server
/02
Movement
  • CharacterMovementComponent vs. Mover decision review — does the choice hold for your combat fidelity target?
  • Mover API churn risk audit (engine version migration cost over the next 12 months)
  • Root motion, prediction, server reconciliation correctness
  • Movement bandwidth — encoded delta budget, prediction window cost
/03
Gameplay Ability System (GAS)
  • GAS extension boundary — what's GAS-native, what's custom, where the boundary should move
  • GAS-tick performance at your density target
  • Predictor design for asymmetric-latency combat
  • Authoring ergonomics — is your design team productive in GAS, or is GAS slowing them down?
/04
Animation pipeline & performance
  • CPU anim-tick budget at peak density
  • Compute-shader pose evaluation — is it on the table, is it worth it for your scene?
  • Anim-graph complexity audit — where the budget is spending
  • LOD strategy — Nanite alignment + skeletal LOD bands
/05
Dedicated server architecture
  • Single-instance scaling profile (CCU per layer; CPU, memory, network ceiling)
  • Sharding, layer / region topology, transitions
  • Backend integration — economy, persistence, matchmaking, presence
  • Build vs. buy review — Pragma, AccelByte, Photon, SpatialOS, custom
/06
Live-ops readiness
  • Telemetry coverage — what you can see, what you can't, what's about to bite you
  • Hotfix / patching architecture — how fast can you ship a server-side change?
  • Observability and on-call posture
  • Incident response and rollback plan

You pick three to five of these surfaces; we calibrate the four weeks to depth-not-breadth.

// 05
PROOF
OF WORK

Three audits we've already done — at our own expense.

These are the three compounding outcomes from the ATONE build that make the audit credible. Each is documented in the post-mortem and provides the methodology and benchmarks the audit applies to your project.

/01
Replication & netcode at 70 CCU per layer. ATONE migrated from Replication Graph to IRIS over seven months. We've published the migration outcome (server tick down ~35%, movement bandwidth down ~60% post-Mover delta codec) and the API churn cost between Epic minor versions. The audit applies the same lens to your stack: where IRIS is paying off, where the migration cost is still ahead of you.
/02
GAS at density without forking the engine. ATONE extends GAS at the edges with custom predictors for Souls-timing combat, while keeping GAS primitives as the authoring surface. The audit answers a question every UE5 multiplayer team eventually has to answer: fork GAS, replace it, or extend it? — with evidence from a project that decided each side of that question, system by system.
/03
GPU-accelerated character animation pipeline. The 70-character ATONE scene cost dropped from 8–10ms of CPU anim-tick to ~2.5ms once pose evaluation moved to compute shaders. The audit applies the same instrumentation methodology to your animation pipeline and tells you whether the same kind of step-change is on the table for your project.

Three tiers. Fixed fee. No T&M.

TierFeeScopeSenior staffing
Audit Light€30,0003 audit surfaces, 4 weeks, 1 lead + 1 seniorLead: senior architect (Lead Engineer tier). Senior: discipline-specialist senior.
Audit Standard€45,0004 audit surfaces, 4 weeks, 1 lead + 1 senior + part-time discipline leadAs Light + part-time Lead Tech Artist or Lead Designer for one cross-cutting surface.
Audit Heavy€60,0005–6 audit surfaces, 4–6 weeks, 1 lead + 2 seniorsTwo seniors run two parallel audit tracks; lead synthesizes.

Fixed fee. Scope is locked at engagement start. If the audit surfaces something that genuinely needs another two weeks to investigate, we'll tell you and quote that as an extension — but the base audit price doesn't move.

// POD FOLLOW-ON CREDIT

If you take a Pod engagement within 90 days of audit delivery, the full audit fee is credited against the first month of the Pod. The audit is the discovery for the engagement; the engagement is the work the audit recommended. Pricing reflects that path.

// 07
THE TEAM

The audit is always senior-led.

The audit is led by a senior architect who has spent five years on a UE5 MMO. The senior staffed alongside the lead is the discipline specialist matched to your audit surfaces.

Audit lead (default): Max Makarenko, CTO & Lead Client Architect. Prior pedigree: Crytek (CryEngine / AAA multiplayer), EPAM Systems (enterprise engineering services), Dragons Lake. Five years on the ATONE backend, IRIS migration, dedicated-server architecture. The author of the technical bets section in our post-mortem.

Discipline seniors (matched to audit surfaces):

  • Replication / netcode / dedicated server — Makarenko's depth area; staffed with a senior backend engineer.
  • GAS / combat / movement — staffed with a senior gameplay engineer who shipped Souls-tier combat in ATONE.
  • Animation pipeline & performance — Lukman Shih (Technical Animation Director) — author of the GPU-anim system.
  • World partition / Nanite / Lumen / level streaming — Serhii Lisetskyi (Lead Tech Artist) — GSC Game World pedigree, S.T.A.L.K.E.R. studio.

The audit is always senior-led. Audit Light has one senior plus the lead — both Senior or Lead-tier engineers. We don't staff audits with mid-tier engineers and we don't subcontract.

Four weeks. Five gates.

W0 (kickoff)90-min discovery call; engagement charter signed; audit-surface list locked; client-side single-point-of-contact named; Perforce / Git access and engagement NDA in place.
W1Lead + senior on-site or remote-attached to your team. Code reading, telemetry review, profiling, interviews with your engineering and production leads. Daily check-in with client SPOC. End-of-week early-findings note.
W2Deep-dive on the highest-risk surface. Reproducer builds. Comparable-project benchmarking against the ATONE numbers. Mid-engagement risk-register draft circulated.
W3Remaining surfaces. Pod-sized engagement spec drafted in parallel — this becomes the optional follow-on path.
W4Audit document final draft → client review → revisions → live readout. Document signed off, deliverables handed over, follow-on path costed and presented.
// 09 FAQ

The questions we get asked.

"We're at 30 CCU today, climbing to 70+. Is the audit still relevant?"
Yes. The audit is calibrated to your target density, not your current density. Most of the architectural decisions that matter at 70 CCU are decisions that need to be right at 30 CCU because they're hard to revisit later.
"We're not on UE5 yet — we're on UE4.27. Can you audit?"
Yes for backend, netcode, GAS, and live-ops surfaces. Mover and Nanite-driven world partition surfaces are UE5-only and not applicable. We'll flag the UE5 migration timeline as a separate deliverable in the audit.
"We're using Pragma / AccelByte / Photon — does the audit cover the middleware seam?"
Yes. We evaluated each of these against custom backend during ATONE; the audit covers what the middleware is doing well, where its limits are for your density target, and the integration seams that need the most attention.
"What if the audit just confirms what we already know?"
Then you have a written, externally-attributed risk register that helps you make the case internally for the work that needs to happen — and a Pod-sized engagement spec that's pre-scoped if you want help executing. The audit is paid; the validation has commercial value even when it surfaces nothing new.
"Who owns the audit document?"
You do. Full client ownership of the audit deliverable. We retain a redacted internal copy for our records (no client-identifying material, no specifics that could leak through reference checks). The post-mortem methodology section we publish publicly — generic engineering lessons, never client specifics — only with your written consent. See our security & IP-handling posture for the full IP-ownership defaults.
"How fast can we start?"
Standard kickoff is 5 business days from engagement charter signed: 1 day NDA + Perforce/Git access; 2 days client-side prep; 1 day GS Studio kickoff prep; 1 day kickoff call. We've started faster when the client's engagement charter is already executed.
"Can we do this under our MSA?"
Yes. We'll mirror your MSA terms or run under our standard MSA — your choice. The audit work product is the SOW.
// 10 START

Book a 30-minute scoping call.

Tell us your engine version, your CCU target, your timeline, and the two surfaces that worry you most. We'll come back with an audit-tier recommendation, a calendar, and a fixed quote within five business days.

Book the call
or scope a longer engagement — Senior-Led Pod →