Khemiri M.Private profile: Upwork shows its earnings only to signed-in members, so we check it every few weeks, not daily. MRR here is a rough estimate (Upwork's six-month earnings ÷ months active). Last read Nov 13, 2026.

⚙️ I build and operate reliable web scraping and data pipelines for AI products, research platforms, and investment workflows. My focus is production infrastructure: systems that keep delivering accurate, structured data across fragmented sources, protected websites, authenticated applications, changing schemas, and multiple jurisdictions. From acquisition to delivery, I turn complex external sources into dependable data your team can use. 📊 Recent production work includes: - 470K+ real estate listings processed daily across nine Swiss sources protected by DataDome, Akamai, Cloudflare, and CAPTCHA. - A US real estate workflow combining Flexmls, Zillow, and WIREdata, with cross-source analysis and automated Google Sheets delivery. - 250+ county and public-record sources integrated. - 70K+ entities extracted from Operabase, followed by IRS Form 990 enrichment for US organizations. - 300K+ retail products monitored every 2–5 hours. - 280K+ automotive parts synchronized daily. - 205K+ business entities extracted and classified. - A 130M-record historical real estate backfill from a DataDome-protected source. - LLM-powered classification, entity extraction, and normalization. 🏛️ I specialize in public records and real estate data. I build acquisition pipelines for court filings, property records, business registries, tax filings, permits, and municipal data. My experience includes Tyler Odyssey, Socrata, Granicus, CivicPlus, ArcGIS, and other state, county, and local platforms. Where jurisdictions share a platform, I reuse the acquisition architecture while adapting schemas, validation, and business logic to each source. These systems support concrete outcomes: tracking ownership changes, identifying distress events, connecting properties to entities, monitoring new filings and permits, and maintaining searchable research datasets. 🔗 I connect fragmented sources into one reliable dataset. Extraction is often just the beginning. I also handle normalization, deduplication, entity resolution, enrichment, and validation. One engagement grew from a paid proof of concept into approximately 30K Operabase profiles and performance records, then expanded into an IRS Form 990 pipeline for US organizations in the dataset. That work included: - Processing multi-gigabyte public datasets and XML filings. - Matching organizations against the IRS exempt-organization register. - Distinguishing affiliated organizations and similarly named legal entities. - Mapping hundreds of financial fields and joining them to the original dataset. - Automatically validating financial totals before delivery. The result: connected, analysis-ready data with quality checks built into the pipeline. 🛠️ I handle the complete data pipeline. Access → Extraction → Normalization → Deduplication → Entity resolution → Validation → Monitoring → Delivery Sources may involve Cloudflare, Akamai, DataDome, PerimeterX/HUMAN, CAPTCHAs, authenticated applications, or undocumented APIs. I choose the most cost-effective, reliable access method for each source and build around the requirements of your production system. Reliability, maintainability, and data quality guide the implementation. ⚡ I make external data usable for AI systems. I add LLM-based classification, extraction, validation, and enrichment where they improve the workflow, then deliver structured results into PostgreSQL, Supabase, Google Sheets, APIs, or your existing stack. I work primarily in Python, using purpose-built infrastructure and keeping orchestration as simple as the project allows. 🎯 We’re a strong fit if: - External data is essential to your product, research, or investment process. - Available APIs do not provide the coverage or detail you need. - Your collectors need to run reliably on an ongoing schedule. - Your sources are protected, authenticated, or spread across many jurisdictions. - Multiple sources need to become one consistent dataset. - Accurate matching, entity resolution, and validation matter. - You want your engineers focused on your product while I manage data acquisition. Engagements can begin with a scoped paid assessment or production pilot, then expand into implementation, additional sources, enrichment, and ongoing operation. 📩 Send me your sources, the data you need, and the product or decision it supports. I’ll give you a candid assessment of the best approach—including whether an existing provider already meets your needs.