Chief Information Officer

Vladimir
Petrov

CIO × AI × ROI

Current — CIO, Gazprombank Tech

I start with project economics. I take responsibility for architecture, the team and bringing the solution into production.

Vladimir Petrov
18+years in IT
up to 170people in teams
> RUB 1bnannual IT budgets
1M+devices in infrastructure
Management system

A living architecture of decisions

From hypothesis to outcome

01

Economics

Objective, impact and payback period

02

Architecture

A system designed to sustain scale

03

Team

Accountability, roles and pace of execution

04

Outcome

A verifiable business metric

CIO · CURRENT ROLE

Gazprombank Tech

Chief Information Officer

IT landscape audit

Audited corporate information systems and IT spend: mapped functionality, support costs, integrations, criticality and system overlaps. Separated verified facts from assumptions and data gaps.

For each system, prepared a decision: retain, optimize, consolidate, replace or retire. Compared vendor operating models using TCO and migration risk, and identified which critical capabilities should be brought back in-house.

InventoryData and costDecision matrixTCO and risk

Telepsychology · B2C/B2B online consultation platform

GRAN.RF

Context

The goal was to launch a commercial online consultation service from scratch: from matching clients with specialists and payments to secure audio and video sessions.

Mandate

CIO and product owner. Accountable for the IT function, MVP and post-MVP roadmap, vendors, infrastructure and information security.

Solution

Built the IT department and engineering capability; launched the MVP in six months and established the platform’s ongoing development.

Outcome

The MVP was delivered on schedule; platform SLA reached 98%.

6 monthsto MVP launch
98%platform SLA
B2C / B2Bplatform business model
CIOand product owner

R&D · UAS · full-cycle manufacturing · AI

Gaskar / View Lab

Context

A technology group developing computer vision products and unmanned aircraft systems. The group included a full-cycle plant — from electronics and component manufacturing to final drone assembly. Scope: 10 functions and approximately 170 people.

Mandate

Accountable for R&D, software development, quality, manufacturing engineering and the AI function. The objective was to connect research, engineering and serial production into one controlled lifecycle.

Solution

Defined stage-gate criteria using ETVX and applied rNPV and PoTS to project selection. Established an independent QA function. In parallel, built the AI function: commercial computer vision products and internal RAG tools.

Outcome

The share of prototypes reaching serial production increased from 30% to 70%. Late critical defects fell by 60%, while first-pass yield reached 99%.

Full cycleelectronics, components and drone assembly
≈170people in teams
30→70%of prototypes reached serial production
−60%late critical defects

The model is only the brain. Architecture delivers the outcome

An LLM alone does not solve a business problem. It needs an AI Harness around it: context, data, memory, agents, tools, orchestration, validation and observability. This architecture turns model capabilities into a system that can be governed, measured and evaluated economically.

Context Engineering · RAG · Memory · Agents · MCP · Orchestration · Guardrails · Evals · Observability

Business problemAI HarnessControlROI