Web scraping & data pipeline contractor

Overview I run a group therapy practice in New York and am launching a therapist directory site. I need a clean, deduplicated database of licensed mental health clinicians in New York State, built in phases by region. I also need the script handed over so I can refresh it later. Phases (each one a separate milestone) Central NY: Onondaga, Tompkins, Cortland, Cayuga, Oswego, Madison, Seneca, Tioga Rochester and Finger Lakes (Monroe, Ontario), plus Buffalo (Erie, Niagara) Capital Region (Albany, Saratoga, Schenectady, Rensselaer), plus Binghamton (Broome) Hudson Valley (Ulster, Dutchess, Orange, Westchester, Rockland) Long Island and NYC (all five boroughs) License types: LCSW, LMSW, LMHC, LMFT, licensed psychologist, PMHNP, licensed creative arts therapist Approved sources only NPPES NPI Registry bulk data (filter by NY and the relevant taxonomy codes) NYSED Office of the Professions online license verification Clinicians’ and group practices’ own public websites, for email and specialties Feel free to use: Psychology Today, Headway, Zocdoc, thrivingcampus.com, professional association directories, listservs, or any login-gated or members-only page. Fields per record First name, last name, credential, license number and status, practice name, street address, city, county, ZIP, phone, website, public email (if listed), specialties (if listed), telehealth (Y/N, if listed), data source, and date collected. Deliverables per phase A Google Sheet or CSV, deduplicated by NPI and license number A short summary: record count, and the percent of records that include an email At the end: a documented, re-runnable script (Python preferred) with setup instructions Requirements Demonstrated experience with NPPES data or other public records Accuracy over volume. I’ll spot-check about 25 records per phase. Screening questions Have you worked with the NPPES NPI bulk file before? What taxonomy codes would you use for these license types? How will you deduplicate clinicians who appear at more than one practice address? Roughly what percent of records do you expect to include an email address, and how would you find them? Budget: based on hourly work