Sebastian Muñoz - Data Engineer | ContraWork by Sebastian Muñoz
Sebastian Muñoz

Sebastian Muñoz

AI-assisted Python automation, CSV tools & web fixes

New to Contra

Sebastian is building their profile!

Cover image for Event replay and deduplication —
Event replay and deduplication — self-initiated demo An AI-assisted Python/SQLite simulation with 16 synthetic inputs processed twice. This is local demonstration work, not a deployed integration or client project. First pass: 8 local requests created, 3 duplicates, 2 conflicting payloads and 3 rejected inputs. Second pass: 0 new requests, 11 duplicates, 2 conflicts and 3 rejections. The database still contains exactly 8 local requests. An intentional failure between the event and outbox writes rolled back without leaving a partial row. SQL checks verified database integrity, matching payloads and no duplicate outbox IDs. Includes source, fixtures, SQLite snapshot, observed results and reproduction commands. No network calls or external messages; this does not claim production delivery guarantees.
0
12
Cover image for Product catalog validator — self-initiated
Product catalog validator — self-initiated demo Built with AI-assisted development using 13 synthetic products. This is an independent demonstration, not paid client work. The Python validator accepts 4 products unchanged and rejects 9 with field-level JSON paths and error codes. Checks include duplicate SKUs, required fields, decimal-string prices, supported currencies, integer stock and HTTPS URL syntax. 11 automated tests passed. The tool does not invent replacements or silently repair records. Source code, sample input, validation report and reproduction commands are included in the demonstration package. Local syntax validation only; no live store integration or network requests.
0
11
Cover image for Inventory reconciliation — self-initiated demo
An
Inventory reconciliation — self-initiated demo An AI-assisted Python tool comparing expected stock with counted stock, using only synthetic data. This is demonstration work, not a client engagement. The executed example retains all 16 input rows for audit, reviews 9 SKUs and finds 3 exact quantity differences. Duplicate SKUs and invalid quantities remain blocked for review; missing counts are not treated as zero. 7 tests passed. Repeated runs produced identical report files. Includes Python source, CSV inputs, JSON/CSV results and instructions. Uses Decimal arithmetic; assumes a consistent unit per SKU. No ERP connection or production deployment.
0
11
Cover image for CSV cleanup and export —
CSV cleanup and export — self-initiated demo Built with AI-assisted development and synthetic data, not client work. A reusable JavaScript module turns 6 sample rows into 4 after removing 2 exact duplicates. Leading-zero IDs and quoted commas are preserved. 16 tests passed, covering quoted fields, embedded newlines, malformed input, optional trimming/deduplication, export escaping and input limits. Includes source code, fixtures, audit JSON and reproducible commands. The image shows actual local results. Exact-row deduplication is not fuzzy identity matching.
0
18