You will get proof your data migrated correctly, checked field by field
Project details
An import that finishes without errors has told you almost nothing. Records get truncated at a field limit, dates shift by a timezone, decimals round, accented names turn into question marks, two real customers merge on a shared email — and the log says success.
I compare the source system against the target, field by field, on a keyed join. Not row counts. Every value.
You get a discrepancy report listing each record that differs, with the source value, the target value and the field, so your team fixes the mapping instead of hunting for the problem. And a plain go / no-go on cutover.
On a sample run of 47,318 records, this surfaced 26 defective records across 6 defect types. The import log had reported zero errors.
Works between any two systems that can export or be queried: CRMs, ERPs, databases, spreadsheets, custom apps. Validation needs read-only access or plain exports — no write access to your production data, ever.
If you have already migrated and suspect something is off, the same check runs after the fact.
Files used only for this work and deleted on request. NDAs welcome.
I compare the source system against the target, field by field, on a keyed join. Not row counts. Every value.
You get a discrepancy report listing each record that differs, with the source value, the target value and the field, so your team fixes the mapping instead of hunting for the problem. And a plain go / no-go on cutover.
On a sample run of 47,318 records, this surfaced 26 defective records across 6 defect types. The import log had reported zero errors.
Works between any two systems that can export or be queried: CRMs, ERPs, databases, spreadsheets, custom apps. Validation needs read-only access or plain exports — no write access to your production data, ever.
If you have already migrated and suspect something is off, the same check runs after the fact.
Files used only for this work and deleted on request. NDAs welcome.
Database Type
MySQL, MS SQL, Oracle, SQLite, PostgreSQL, MongoDB, Couchbase, Realm Database, Azure Cosmos DB, LevelDBWhat's included
| Service Tiers |
Starter
$190
|
Standard
$430
|
Advanced
$780
|
|---|---|---|---|
| Delivery Time | 3 days | 5 days | 8 days |
Number of Revisions | 1 | 2 | 3 |
Source Code | - | - |
Optional add-ons
You can add these on the next page.
Fast Delivery
+$110
Extra table or object
(+ 1 Day)
+$95
Post-cutover re-validation run
(+ 2 Days)
+$140
Ongoing load checks + 30 days of support
(+ 2 Days)
+$210Frequently asked questions
About Evgenii
Document Data Extraction & Reconciliation | PDF to Excel/API | Python
Praha-Nove Mesto, Czech Republic - 5:58 pm local time
pay applications, statements, EOB-style documents) into clean, reconciled data —
and tell you exactly which line items don't match your contract or ledger.
Most extraction gigs stop at "here's your Excel file." That's where errors hide.
My pipelines add a reconciliation layer: every extracted number is checked
against an external source of truth — a signed contract, a rate sheet, a PO,
a prior-period ledger — and mismatches are flagged with evidence, not buried.
What I do:
• PDF/scan → structured data (tables, line items, multi-page, dirty scans — OCR + LLM + rule-based validation)
• Invoice & AP reconciliation: 3-way match (PO / receipt / invoice), duplicate detection, price-vs-contract audit
• Construction pay applications: progress billing vs. schedule of values, change orders, retainage math
• Recurring pipelines: email-in → parsed → validated → pushed to your sheet, QuickBooks, or API (cron, webhooks)
• Human-in-the-loop review UIs when "errors are unacceptable" — you confirm flagged fields, not re-type documents
How I work:
• Fixed accuracy targets agreed upfront; I report precision on a sample batch before you commit
• Your data stays yours: processing can run in your cloud account; NDAs welcome
• Zero-dependency, documented code you own — no lock-in to my tooling
Stack: Python (pdfplumber, OCR), LLM extraction with schema validation,
Supabase/Postgres, serverless cron (Modal), Stripe for metered pipelines.
If your team spends hours re-typing documents — or worse, trusts numbers
nobody verified — send me one sample file and the source of truth it should
match. I'll return a parsed + reconciled sample within 24 hours, free.
Steps for completing your project
After purchasing the project, send requirements so Evgenii can start the project.
Delivery time starts when Evgenii receives requirements from you.
Evgenii works on your project following the steps below.
Revisions may occur after the delivery date.
1. Scope confirmation and sample review
I look at a sample export from both systems, confirm there is a stable key to join on and agree which fields matter. If a meaningful comparison is not possible, you hear that before any work starts.
2. Free sample comparison: 500 records
I compare five hundred records across both systems and send the result, so you see the defect classes it catches before committing to the full run.