Denis Nagaev

@nagaev

Co-founder and CTO in AI/ML agency

In ML and LLM since 2019. Shipped RAG assistants and forecasting models to production for enterprise clients across industry and retail

Open to

Funding & grantsProject collaborationMentoring othersCompany partnershipsCommunity initiatives

Career

Experience: 7 years

Experience

Chief Technology Officer

OSMI-IT LLC

AI systems for large enterprise clients. Around 120M RUB annual revenue, 40 staff, 15 recurring clients, among the twelve largest Russian firms in this niche.

Nov 2019 — Present

Built and leads OSMI-IT's technical organization and enterprise AI delivery; grew average project value to 4M RUB and led production systems with measurable client outcomes.

• Took the company through three shapes: web studio, custom development contractor, platform vendor. Average project value rose from ten thousand rubles to 4M RUB. • Grew divisional revenue from 52M to 100M RUB across 2025 and 2026. • Owned the technical half of pre-sales: around 120 proposals, budgets from 1.9M to 18.5M RUB, and 60M RUB in closed deals; checked estimates for realism before commitments reached clients. • Made payback analysis mandatory before work starts, covering cost per model operation and time to return. Projects that do not pay back are stopped at the gate.

Education

MSc, Applied Mathematics and Informatics, program in AI Research and Entrepreneurship

HSE University, Faculty of Computer Science

2024 — 2026

MSc, Machine Learning

ITMO University

2018 — 2020

BSc, Fundamental Informatics and Information Technology

Samara National Research University

2014 — 2018

Selected projects

Avito — procurement master data

Led the process audit, normalization rules and selection of the target system and machine-learning approach for deduplicating around 3M inconsistent records across thirteen systems. Reduced these to 4,500 canonical items; reported 33M RUB saved on inventory management in 2026, a further 12M RUB in avoided purchases, and five roles freed up. Role: project CTO.

FSK Group — pre-litigation claims from residents

Built a workflow for document recognition, defect classification, reconciliation of reports against expert assessments, cost calculation and export of claim cards. Review time fell from 40 to 12 days, coverage rose from 60 to 100%, F1 exceeded 0.90, and uptime was 99%. Role: QA lead.

Erich Krause — demand forecasting

Forecasting across 11,000 SKUs with daily retraining, stockout and dead-stock dashboards, and accounting-system exchange. Forecast accuracy reached 91%, on-shelf stockouts fell 28%, and uptime exceeded 99%. Shortlisted for the Workspace Digital Awards. Role: project CTO.

Skolkovo Foundation — single entry point to support services

Designed the architecture and owned the corporate-content pipeline to a vector store, FAQ-service exchange and knowledge-base administration model. Answer accuracy was 87.8% against a 70% acceptance threshold; answers were delivered within ten seconds and handled about 70% of queries. Compared five language models across twenty test scenarios.

ROSMEN — customer review handling

Combined marketplace and website reviews into one workflow with sentiment and severity scoring, drafted replies and routing. Manual triage fell to 10% or less, negative-review response time fell from seven days to one, reply coverage rose from 30 to 95%, and returns fell 9%. Role: project CTO, team of seven.

Zaslon — component selection from plain-language queries

Built a system that weighs price, lead time, availability and application conditions and proposes substitutes. 87% of queries are resolved without a human, selection time fell to two to four seconds, and staff without engineering training can use it. Role: project CTO.

Achievements

AI Startup of the Year category

PRIORITY: DIGITAL · 2026

Top ten in conversational assistants and top fifteen in AI

Runet Rating · 2026

Skills

PythonFastAPIDjangoasyncioCeleryTypeScriptNode.jsJavaC#PostgreSQLpgvectorClickHouseRedisKafkaRabbitMQBullMQS3-compatible storageKubernetesDockerHelm