THE MAP · 4 STATIONS · CLICK TO TRAVEL

Artem Baikov

CX Analytics Lead · Senior Data Analyst

I turn customer feedback into metrics that drive business decisions.

Cádiz, Spain · Authorized to work in Spain · Open to remote
English · Español · Русский
artem@artembaikov.com LinkedIn Download CV (PDF)
~1,000
supermarkets with bonuses based on my index
~20k
reviews tagged by an LLM every week
~50
regular reports, data collection fully automated
6
years in customer experience analytics

About

For 6 years I have worked in customer experience analytics at X5 Group, Russia's largest grocery retailer. I lead a team of 3 analysts. We run about 50 regular reports and dashboards, from single stores up to the Board of Directors.

Case studies

01

Customer Happiness Index

One score for each store, built from complaints, online map reviews and surveys. Managers can go down from division to cluster and to each store.

Basis of bonuses for ~1,000 supermarkets for 3 years
Read more
  • Three feedback channels, each with its own weight, plus rules that set the score to zero in certain cases.
  • One number for each level of the chain: store → cluster → division.
  • It was the basis of store bonuses across the whole chain for 3 years.
KPI designPythonSQLQlik Sense
02

Review classification with LLMs

An LLM pipeline puts tags on reviews about product quality: about 20 categories, several per review. It replaced my older rule-based classifier.

~20,000 reviews a week · checked against a second model and by hand
Read more
  • I checked quality in two ways: about 95% agreement with a second, independent model, and a manual sample of reviews for every tag.
  • I improved the prompt using its mistakes, for example by separating “expires soon” from “already expired”.
  • The tags separate problems that the order picker controls from unclear ones. This made a new bonus and penalty system for pickers possible.
LLM APIsPrompt designText classificationDashboards
03

Conjuga.pro — a product I built and use

An app to practise Spanish verb conjugation. I made it to learn myself, and it is on Google Play. Every answer is saved in PostgreSQL, so it is also my product analytics lab.

Daily practice since May 2026 · 6,000+ saved answers
Product analyticsA/B testingSupabase / PostgreSQLPWA

Read the case study →conjuga.pro ↗

04

LeetCode practice with my own ETL

Mostly Python, plus 320+ SQL and pandas problems; two thirds of my Hard problems are SQL. My Python ETL loads my progress from the LeetCode API into SQLite and feeds a dashboard.

1,500+ problems · 1,300+ days of practice
SQLPythonETLChart.js

See the dashboard →

How I work

AI in my workflow

LLMs in analyticsBesides the review tagging in case 02, I use LLMs to group texts (clustering) in exploratory analysis.
Local AI modelsI compared three open-source voice cloning models on my own GPU, trained them on 27 minutes of my voice, and automated video making for the programming lessons I create for my son.
AI agents every dayI use coding agents to build pipelines, set up environments and check results. The decisions and the final checks are mine.

Skills

Analytics: KPI and index design · CX metrics (NPS, CSAT) · A/B testing · text classification and clustering · LLM labeling · executive dashboards

Tools: Python (pandas) · SQL (Greenplum, PostgreSQL, Hive) · Qlik Sense · Tableau · Looker Studio · Supabase · now learning dbt and Airflow