CTO / CPO · Enterprise AI

Enterprise-grade AI systems,

built and shipped at national scale.

30M+ users · 95+ awards · 17+ years of enterprise experience

View Track Record

Contact

MERCEDES-BENZ·NLMK·CATERPILLAR·NORNICKEL·GTLK·

Built by our founder, in production at these organisations

Products

Products you can download and run

Local-first desktop apps. The model runs on your machine, so the data stays there.

Askbase

Ask your database in plain words

A macOS app that answers questions about your own database and shows the SQL behind the answer. The model is a file on your disk, so the question and the data never leave the machine. Works offline, no account.

Built on the on-prem Text-to-SQL research (arXiv:2606.29733)

macOSLocal modelText-to-SQL

Open

pii-roundtrip

Redact documents before ChatGPT, offline

A macOS app that swaps names, NHS and National Insurance numbers, bank details and addresses in Word and PDF files for placeholders, then restores them in the AI answer. 994 of 1,000 test documents came out with no personal data left.

macOSOpen sourceApache 2.0

Open

AI Ecosystem

Technology behind the systems

AI Systems

NLP & LLM

Citizen appeal classification and response generation, in federal production

Computer Vision

Document and image recognition in appeal processing

ML Ops

Retraining loops on operator-verified data, production monitoring at national scale

GenAI

Text-to-SQL and AI assistants in the BI+AI management platform

Big Data

Terabyte-scale receipt stream for the national Consumer Price Index

Neural Networks

Word2Vec+LSTM and PyTorch classifiers in federal production

Select a domain for details

30M+

Users impacted

95+

Industry awards

15+

Industries served

AI systems and platforms built and shipped across 15+ industries — at national scale.

Systems in production

Systems that shaped industries

In production · Ministry of Construction

AI-powered citizen request system

AIGovernment2023–2025

78% in research → 95%+ accuracy in production · Rospatent reg. 2026662502

Outcome

XXVI Golden Site Awards, 1st place 'Best AI-based solution for citizen-inquiry processing' (2023, awarded 2025)

In production · Russian Railways

RZD Bonus Loyalty Platform — Highload

EnterpriseLoyalty2020–2023

33M passengers, 9.9M members

Outcome

XXVI Golden Site Awards, 1st place 'Best Corporate Programme' (2023, awarded 2025); Tagline 'Best Integration' 2020-21, 'Best Highload' 2022

In production · Federal Statistics Service

AI-powered Big Data system for Consumer Price Index

AI & DataGovernment2021

Terabyte-scale data, 90% accuracy in production, 150+ report templates · Rospatent reg. 2026662475

Outcome

Tagline 'Best Audience & Data' 2022 (1st place), 2023, 2024

In production · Mercedes-Benz Bank

Online credit & leasing application

FinTechEnterprise2022

Outcome

Tagline Award 2022

In production · State Transport Leasing Company + STLC Europe

International corporate omnichannel platform

EnterpriseInternational2020–2023

Outcome

Tagline 'Best International Site' 2020-21; 'Best Integration' 2023; 'Best International Digital Service' 2025

In production · DigitalCEO

BI+AI digital management ecosystem

AI ProductAI Engineering2025

Own product · #3 BI vendor by licence revenue (TAdviser)

Outcome

CNews AI Industry Review 2024

In production · NLMK

Corporate portal for 60,000 employees

EnterpriseIntranet2019

Outcome

Tagline 'Best Intranet' 2019; StepTwo Intranet & Digital Workplace Awards — 2nd place worldwide (Silver)

In production · Federal Alcohol Market Regulation Service

AI-powered citizen request processing system

AIGovernment2022–2023

Same registered product · Rospatent reg. 2026662502

Outcome

Tagline 'Best AI Technology' 2022 (Bronze); 'Best Gov Service' 2023

In production · Sberbank

Cybersecurity education platform

EnterpriseCybersecurity2021

Outcome

Automated cyber literacy system

In production · Caterpillar Financial

Online leasing applications

FinTechEnterprise2020–2021

Outcome

Tagline Award 'People's Choice' 2020–2021

Research

Peer-reviewed AI research

Four sole-authored papers on arXiv (DOI). Indexed in NASA ADS, Semantic Scholar & Google Scholar.

ORCID 0009-0004-7005-6242

arXiv

AI Appeals Processor: A Deep Learning Approach to Automated Classification of Citizen Appeals in Government Services

Behind the citizen-request AI (Ministry of Construction, Rosalkogol)

arXiv:2604.03672 DOI: 10.5281/zenodo.19393274

arXiv

Machine Learning for Coding Retail Product Names to Consumer-Price Categories: A Rule-plus-Bag-of-Words Pipeline with Reliability-Weighted Human-in-the-Loop Labeling

Behind the CPI big-data system (Rosstat)

arXiv:2606.02004 DOI: 10.5281/zenodo.20503356

arXiv

How Far Do On-Prem Open LLMs Get on Text-to-SQL? A Cross-Family Size × Technique Frontier on BIRD

Behind BI+AI / Text-to-SQL

arXiv:2606.29733 DOI: 10.5281/zenodo.20952794

Speaking

Conference talks

Conf42 AI Agents 2026

Agentic Text-to-SQL in Production: What 500+ Data Marts and 30M Users Taught Us About LLM Orchestration

On running on-prem NL2SQL with open-weight LLMs in production at national scale

Peer review & judging

Conference program committees and awards judging, 2026

Program Committee Reviewer

NeurIPS 2026 Workshop on Continual Learning for Enterprise AI Agents (CLEA)

Reviewed submissions on continual learning, agent memory and enterprise agent systems

Program Committee Member (Reviewer)

NeurIPS 2026 First Workshop on Reliable Evaluation for Language Models (JUDGe)

Program Committee Member (Reviewer)

NeurIPS 2026 Workshop on Lightweight Trustworthy Foundation Models (LIGHT)

Reviewer

2026 IEEE 4th International Conference on Digital Transformation: Algorithms, Computing and Future Technologies (Al-Khwarizmi)

Citizen-appeal classification (NLP) · Khiva, Uzbekistan

Reviewer

2026 IEEE 4th International Conference on Problems of Informatics, Electronics and Radio Engineering (PIERE)

Computational linguistics and NLP section

Judge

2027 IMPACT Awards (Simpler Media Group)

In the media

Expert commentary in the media

Techrecast

Interview on enterprise AI: the decision, not the model, is the right object

Full solo interview, cited as "founder and CTO of Besk Tech"

ITWeb Africa

On vendor lock-in as technology's hidden tax

Expert commentary, cited as "chief technology officer of Besk Tech"

TechRound

On 'superintelligence' and the AI industry's real problems

Expert commentary, cited as "Enterprise AI Architect, CTO and Founder at Besk Tech"

Founder

Vladimir Beskorovainyi

CTO / CPO · Enterprise AI Systems

Vladimir Beskorovainyi builds production AI and big-data systems at national scale. Over 17+ years he has led engineering for platforms serving tens of millions of users: from NLP systems classifying 50M+ citizen requests to big-data pipelines behind national statistics.

Full lifecycle: architecture, machine learning, and shipping models to reliable production. Products adopted by federal agencies and enterprises, with 95+ industry awards.

Founder's Portfolio

Contact

For collaboration, technical discussion, or speaking — reach out directly.

admin@besk.tech LinkedIn Telegram