Build a Python quality gate that separates genuinely thin and doorway pages from thin-looking pages that rank, using cheap structural signals first and an LLM judge only on the ambiguous middle.
Build a Python pipeline that extracts entities with spaCy and Google Cloud NLP, computes an entity gap against the SERP, and turns salience scores into content decisions.
Approximate Google’s Quality Rater Guidelines at scale: a Python pipeline with a QRG-mapped rubric, an evidence-bound LLM prompt, kappa calibration, and a refresh queue ranked by quality gap times traffic.
A Python audit that harvests parameterized URLs, classifies them into pagination, sort, and filter buckets, checks canonical and robots handling, and ranks the crawl traps inflating your index.
A Python pipeline to audit canonical tags at scale: catch conflicting signals, canonical chains, non-indexable targets, normalization mismatches and sitemap disagreements before they cost you rankings.
Build a Python keyword research pipeline that harvests seeds, expands with Google Autocomplete, mines People Also Ask, and scores opportunity on your own terms.
A weekly Python pipeline that extracts every hreflang annotation, verifies reciprocity, validates language codes, and checks target health — alerting only when a cluster actually breaks.
Detect real keyword cannibalization from Search Console query-and-page data with a Python pipeline that measures impression share and week-to-week position flips, then routes each case to consolidate, differentiate, or demote.
Aggregate traffic hides per-URL content decay. Build a Python pipeline that pulls 16 months of Search Console data, uses the Mann-Kendall test and Theil-Sen slope to tell real decline from seasonality and noise, and outputs a ranked refresh queue.
Embeddings measure meaning; near-duplicate pages are about shared text. Here is the MinHash + LSH pipeline in Python that catches variants, boilerplate, and templated thin content at scale — and a decision framework for what to consolidate, canonicalize, or prune.