A full analysis pipeline that turns a dataset into a finished report, not just charts. Hand it a CSV, Excel, Parquet, or JSON file and it cleans the data, scores its quality, runs EDA, and where the data supports it, trains a prediction model with SHAP attribution and a backtested forecast. You get a trust score, ranked findings, and recommendations as a Word doc, an interactive HTML report, and an optional PowerPoint deck. The numbers come from code, so they stay consistent, never invented.
Built from practical B2B sales research workflows for founders, agencies, consultants, and sales teams. This Skill turns visible company website evidence into ICP fit scoring, buying signals, pain hypotheses, buyer recommendations, and outreach angles so users can personalize outbound without making unsupported claims.
I build and run the web-scraping engine behind a production AI assistant. The hardest sites on the web — JavaScript-heavy apps, empty sitemaps, and content behind search forms — hand a normal crawler nothing but cookie banners. Over months I learned to make them give up clean data: render fully, find the hidden pages by search, dedup by stable IDs, and quality-gate every page so junk never reaches your dataset. I've distilled that pipeline into this Skill — self-healing and cost-capped.
Local-run B2B lead generation for LinkedIn. Search ~20-30 profiles per run, cross-verify named profiles, enrich anonymous ("LinkedIn Member") entries via company team pages, export a confidence-labeled Excel. Runs on the buyer's own machine with their own browser and their own LinkedIn login. Designed for B2B sales teams, agencies, and solo founders running outbound.
Turn any structured dataset into a full programmatic-SEO site: hundreds of auto-generated, indexable landing pages built around the recurring entities in your data, complete with sitemap, robots.txt, and a GDPR/AdSense-ready privacy policy.
Turn underperforming listings into converters: titles, bullets, images, keywords, and review-mining workflows.
将企业文档整理成可追溯的 JSON 知识库，自动提取事实、合并重复记录、标记冲突，并保留原文出处。
Check HTML files for the markup problems that most often slip into hand-written pages, templates and email markup: unclosed and mismatched tags, duplicate ids, images without alt text, missing or empty titles, a missing lang attribute or doctype, deprecated tags, and links without a target. Runs fully offline with Python's standard library, reports every finding with a file and line number, and never copies attribute values into its reports.