CASE · ANALYTICS
An analytics and reporting platform that automatically pulls marketplace data, bank statements and cost prices into a single model and gives the seller cash flow, P&L and per-SKU unit economics — without manually merging dozens of Excel files. Built for a single client and deployed inside their infrastructure.
THE CHALLENGE
The client is a seller on the Wildberries and Ozon marketplaces who was drowning in scattered data: platform reports, bank statements, cost prices, commissions, logistics, advertising — all in different formats and tables. Understanding real profit per product, forecasting sales and seeing cash movement without manually merging dozens of Excel files was practically impossible.
The goal was a single platform built around their workflows that:
FEATURES
A journal of operations (income, expenses, transfers), mapping to categories and accounts, automatic classification. Cash flow and P&L reports are built from primary records.
“Sales by product”, “Unit economics” and “Sales plan control” reports — with margin, drill-down to a single operation, sparklines and trend charts.
Cost price tracking by date and automatic allocation across sales for correct profit calculation.
Monthly sales plans per product with import, copying and a change log.
Import of bank statements and arbitrary Excel/CSV files with configurable column mapping and templates.
Permission groups, user invitations, granular access to data and reports.
A knowledge base and a news feed for platform users.
ARCHITECTURE
The system is built as a set of independent services rather than a monolith — so heavy data synchronization never slows down the UI.
The public API for the frontend and all business logic. Runs in cluster mode and scales across CPU cores.
A separate sync process: pulls data from the external marketplace database (PostgreSQL), normalizes it and lays it out into pre-aggregated data marts in MySQL. Cron schedulers, a mart-recalculation queue, automatic backups.
WebSocket notifications: sync progress, toasts, menu and page updates without reloading.
Serves the SPA and auto-refreshes open tabs on deploy (SSE + build version check).
Groundwork for the future: the user asks a question in natural language, the service generates and executes SQL over the data via an LLM and streams the answer.
Database access goes only through stored procedures and pre-aggregated data marts. This keeps reports fast on large volumes and gives a single point of control for business logic on the database side.
INTERFACE
ENGINEERING CHALLENGES
This is not CRUD — most of the work went into data correctness and performance:
STACK
HTTPS communication between services with service-level authorization; MySQL access goes only through sp_* stored procedures.
A working production platform that replaces manual report merging and gives the seller a realtime financial picture of the business: from cash flow and P&L to per-SKU margin and sales plan control. Designed, built and shipped to production by one person — including the database schema, every backend service, the entire frontend and the deployment infrastructure.
SEE ALSO
Management accounting and sales analytics, 500M+ transactions, calculation core inside the database.
A morning digest, anomaly detection, unit economics and restocking. Code does the maths; the agent only reads.
Service pages with scope and pricing are available in Russian.
CONTACT
I'll automate data collection and financial reporting for your business — from architecture to production. Message me on Telegram — we'll scope the task, timeline and budget.
Message on Telegram dnovoselov@bk.ruI reply within 3 hours on weekdays
A project's budget and commercial terms belong to the client, and I don't publish them. Your project will be treated the same way. My own price ranges are in the Pricing section.