Async · Remote · Worldwide
Data quality engineering — audit, cleanup, import preparation.
I find what's wrong in your data before it reaches your dashboard.
Numbers look off but nobody has time to investigate. I find the root cause and document it.
CSV/Excel files that fail on import to Odoo, WooCommerce, HubSpot, or CRM. I clean and validate before they go in.
External data sources delivering files with structural anomalies. I segment by vendor and surface the patterns.
Dataset too large for the team to audit manually. I handle scale — 1M to 50M+ records.
Case studies
48.7M records
26.6% anomaly rate. Vendor-specific financial risk isolated — 5.8% of Vendor 2 trips contained negative monetary values, pointing to a structural issue, not random noise. DST-aware timestamp logic prevents false deletion of valid records.
Open on GitHub →1M records
Placeholder ratings (~259K records) and invalid geographic values isolated before category benchmarking and location comparisons. Demonstrates how data can look complete while producing systematically wrong conclusions.
Open on GitHub →2M+ records
Time-series cleaning — datetime reconstruction from split fields, incomplete row identification and removal, and preparation of a validated dataset for reliable hourly, daily, and seasonal usage analysis.
Open on GitHub →In review
Relational data quality audit using SQL — validation logic, anomaly detection, and structured output for downstream reporting. Publishing shortly.
GitHub →Send a sample file or describe the dataset. I reply within 24 hours with a direct answer on scope and timeline.
Every flagged record stays traceable — nothing disappears without an audit trail. You get a cleaned file + a full change log explaining every decision.
A working public tool — OBD2 fault code reference for Polish-speaking car owners. Built from scratch: 100/100 Lighthouse scores across performance, SEO, accessibility, and best practices. Zero cookies, zero tracking, zero ads.
Not data engineering — but demonstrates what "built properly" looks like end to end.
Open site →Best fit: data audits, cleanup pipelines, import preparation, anomaly isolation, validation before dashboarding or reporting. Async. No calls required to get a quote.