28 projects. All production.

Real-time, unified feed across Yad2, Madlan, Homeless and Facebook - AI-normalized in Hebrew.

Aggregated jobs from Workday, Comeet, Greenhouse and Lever with a Hebrew-first RTL UI, agency portal and real-time chat.

Three-module automation: PDF questionnaires parser, Rise4→Monday→CharmHealth sync, eClinical reconciler.

Tablet counting UI + Flask API + Playwright/Henry Schein cart automation - end-to-end with documented handover.

Daily Spanish-news pipeline: RSS → LLM rewrite + verify → optimized images → GitHub PR.

Self-hosted DocuSign alternative with proper Hebrew rendering - visual field placement, signer link, pdf-lib + fontkit fill.

Full RTL Hebrew bakery platform - public storefront, custom-cake flow, and 12-page admin CMS.

Gmail → Worldfilia CRM match, per-UTM invalid-rate scorecards to Google Sheets.

End-to-end Dell.com add-to-cart + guest checkout automation with anti-bot humanization and failure artifacts.

Aggregates UK-sponsor jobs from HireMyVisa, enriches sponsor status in batches, publishes a sales-ready Excel workbook.

Apps Script scraper + branded, password-less client dashboard - one deploy, per-client URL.

SERP + Yelp + PsychologyToday scraping → website/email enrichment → quality analyzer → batched SMTP outreach.

Python cleaner that normalizes a messy legacy referrals workbook into a migration-ready template.

Dual-surface (sheet modal + web app) data manager with column whitelists for safe partial editing.

Audited a failing LoRA run, fixed 4 real bugs (including an Unsloth MoE naming trap), shipped a handover bundle with honest data-size expectations.

Inherited Django e-commerce app - stabilized, re-documented, editable-install fixes, DB backup fixture for fast onboarding.

Short-engagement debugging: reproduce the CI-flaky Playwright spec, fix it, hand back a minimal diff.

Two Android apps (notification-listener + accessibility-service) that push WhatsApp keyword hits to a webhook.

8 scrapers across 4 countries (Chile, Mexico, Puerto Rico, Spain) feeding two unified JSON schemas.

Adaptive partitioning conquers a 150-record server cap to deliver every offender across 323 Georgia institutions.

Cloudflare bypass + Next.js __NEXT_DATA__ extraction + API chaining to pull full Crexi CRE details.

Cookie-auth Selenium scraper that pulls latest N posts/reels from FB profiles with atomic writes and checkpoint detection.

10+ PA civil + public-notice portals into a unified keyword-categorized legal-leads pipeline with BatchData enrichment.

Walks codes.iccsafe.org's premium viewer and exports every section to a single merged PDF via Playwright storage_state.

Turn-key ZIPped Python scraper + PDF walkthroughs + DOCX output template - handover for a non-developer team.

Headless Python tool that pulls Instagram profiles, posts and Reels via internal REST + GraphQL APIs.

Single-CLI Python scraper that bulk-pages the DLCP paginationadvance-search API.

curl_cffi + browser-impersonation scraper that dumps the full Israeli KSP catalog to JSON + Excel-friendly CSV.