Alexander Bakharev

Alexander Bakharev

I take AI from stalled pilot to a production line with a metric. ×2–3 content output, same team · cost per asset −40–60% · Remote EU

AI production operator, not advisor. 15+ years running content production in game development — teams to 150+, €6M+ P&L — the last three spent removing millions in production costs with AI pipelines that run on real workloads. If your studio or agency already bought the tools and the results never arrived, that's the problem I fix.

Sound familiar?

"The pilot was impressive. It died the day it met our real process."

Behind it: nobody wired the pilot into roles, QA, and tooling — demos don't survive contact with production. → Production Pipeline

"We cut people to run on AI. Now production rests on two enthusiasts and luck."

Behind it: the process lives in heads, not in playbooks — one resignation away from a standstill. → Production Pipeline

"We pay for every AI tool on the market — and output hasn't moved."

Behind it: no baseline was ever measured, so nobody can see where the money leaks. → Process Diagnostic

"AI generates fast. Review eats all the savings."

Behind it: generation scaled, quality control didn't — QA gates are the missing half of the pipeline. → Production Pipeline

"Everyone on the team uses AI their own way. Quality is a lottery."

Behind it: personal hacks instead of a designed pipeline with roles and standards. → Production Pipeline

"Clients demand AI prices — and I don't even know our real cost per asset."

Behind it: no per-unit economics. You can't defend a price you can't compute. → Process Diagnostic

None of these are tool problems. They are production problems.

MIT: 95% of GenAI pilots never reach production. S&P Global: 42% of companies abandoned most AI initiatives in 2025. If the list above reads familiar, you are the market norm — and the fix is process design, not another subscription.

How I Fix It

Process Diagnostic

€4,000 fixed · 2 weeks

One production process — not a company audit. You get a launch-ready solution: the current cost and speed of the process in numbers, the target architecture (agents, tools, QA gates), an implementation plan, and the ROI math. Execute it in-house or with me.

Credited in full against a Production Pipeline engagement.

Production Pipeline

from €20,000 · 4–8 weeks

A content production line in operation — art, copy, localization, ASO, or LiveOps content. Inside: throughput ×2–3 with the same team, QA gates instead of manual review, one quality standard instead of personal hacks, and a live cost-per-asset panel. Your team runs it; I hand over playbooks, not dependencies.

Milestone payments 30 / 40 / 30 with an exit point at every stage · One pipeline project at a time.

Ongoing support available after delivery.

I Don't Take the Project If

Describe one production process in one email. Within 48 hours I'll tell you whether there's money in it — and roughly how much.

Send a process for triage

Free. No deck, no discovery call required.

Proof

−€3.1M+ removed · 2 pipelines
10M+ items in AI pipeline
×3 division output
€6M+ P&L managed
150+ people led
70+ shipped titles

−€3.1M+ breaks down into two shipped cases: €2.4M (generative 3D pipeline) + €700K/year (in-house image-generation pipeline) — both below.

Generative 3D Pipeline — 10M+ Item Catalogue

Production cost cut from ~€2.8M to €400K on the second iteration: concept generation, image-to-3D, procedural multiplication, automated QA gates, headless rendering — with licensing and IP clearance as architecture constraints.

Before / after: ~€2.8M baseline → €400K delivered.

In-House Image-Generation Pipeline for Game Art

Custom-trained pipeline embedded into a live studio's art production — external art costs eliminated, concept iteration accelerated.

Before / after: −€700K/year external art spend · concept iteration −40% time.

MMO/RPG Division Rebuilt

Team and process redesigned from the ground up — same headcount, restructured production system.

Before / after: ×3 output · projected 3-month slip compressed to 2 weeks.

Full case library with stacks and methods — AI Cases →

Other Practices

For Game Studios

Content pipelines and production leadership in studio terms — LiveOps content, art production, cost per asset. 15+ years, MOBA to hybrid-casual.

Studio page →

OSINT & Due Diligence — Escherlock

Independent due-diligence practice: counterparty risk, UBO verification, asset tracing, subject investigations. Decision-grade, lawfully sourced.

About the practice →

Certifications

Anthropic — April 2026 Building with the Claude API · AI Fluency: Framework & Foundations · Introduction to Agent Skills · Claude Code in Action · Introduction to MCP · MCP: Advanced Topics · Claude 101
Other Google Project Management · Google Prompting Essentials · Generative AI for Cybersecurity Professionals (IBM) · AI in Education (UPenn) · Gamification (UPenn) · Game Theory (Stanford)
LinkedIn GitHub Crunchbase