Skip to content
AS

Artem Sidorov

@ArtCoreMLConnected with DSML. This person's Telegram account has been confirmed through DSML Kazakhstan.

Senior AI / NLP Engineer (6+ Years of Experience)

Senior AI / NLP Learning Engineer with 6+ years of experience building production NLP and applied AI systems for customer support, knowledge search, analytics, and trust & safety.

Open to

Research collaborationCompany partnershipsSpeaking & workshopsStartup collaborationHiring & referrals

About

Owns the full lifecycle from data and evaluation to deployment, staged rollout, and monitoring, with a recent focus on RAG, LLM agents, and Text-to-SQL.

Career

Experience: 6 years

Experience

Senior AI / NLP Learning Engineer

XSOLLA

A global video game commerce company that provides payment, publishing, and related services to game developers and players.

Jul 2023 — Present

Built production support RAG, multi-agent, and Text-to-SQL systems, with reported gains in retrieval quality, automation, and support workload.

• Partnered with Support leadership to define quality, safety, and latency targets for a multilingual, on-premises customer-support RAG platform; guided architecture and staged production rollout. • Designed hybrid retrieval (Qdrant, Elasticsearch/BM25, RRF), domain-tuned embeddings, cross-encoder reranking, and ACL/version-aware indexing. Built real-query evaluation and citation checks; achieved 93% Recall@10, 91% nDCG@10, 93% faithfulness, and 94% citation support.

Machine Learning Engineer

inDrive

A global mobility and urban services platform best known for ride-hailing, connecting passengers and drivers who can agree on fares.

Apr 2020 — Jun 2023

Delivered an employee knowledge assistant, multilingual ticket classification, and high-volume moderation, with measured gains in search speed and automated handling.

• Led ML work on a bilingual internal knowledge assistant for ~1,200 employees, progressing from retrieval-only search to approved LLM-generated answers in 2023. Improved Recall@10 from 82% to 91% and nDCG@10 by 7 points; observed search time fall from ~90s to 10–15s. • Built multilingual support-ticket classification from ~500K historical records; worked with domain experts to refine a 15-class taxonomy and reannotate 15K examples. Achieved 0.89 macro-F1 and 85% auto-routing coverage, with corrections on automated tickets down to ~9%.

Education

Master, Data Science

Nazarbayev University

2022 — 2024

Bachelor, Computer Science

Moscow Institute of Physics and Technology (MIPT)

2017 — 2021

Skills

vLLMAzure OpenAIFastAPIREST APIDockerKubernetesAirflowRedisRabbitMQPostgreSQLRAGagentic RAGAI agentsReActtool/function callingLangChainLangGraphText-to-SQLprompt engineeringstructured output