AI Coding Engineer (Claude Code): Search & Shopper Experience for Auto Parts E-commerce

*LONG TERM OPPORTUNITY* *MUST WORK USA HOURS NEW YORK TIME ZONE 830 AM TO 5 PM - NO EXCEPTIONS* *MUST USE CLAUDE CODE MAX - VERY IMPORTANT!* We're an auto parts e-commerce retailer with 2M+ parts from 150+ brands. We want one AI coding engineer who owns everything the shopper touches, from the first Google or ChatGPT answer that brings them in, to search, to the product page, to the email that brings them back. Most of all, you'll own search, because that's where every buyer starts and ours isn't good enough yet. This is an AI-first engineering role: you'll build mostly with Claude Code (agents, specs, reviewing AI output). We don't care how many lines you type. We care how fast you ship things that work, and whether you can prove it with numbers. WHAT YOU'LL OWN 1. Search (about half your time, and the top priority) - The whole search journey: search box, suggestions, results, filters, product page, on phone and desktop - The engine: our catalog in Typesense (or similar), kept in sync with price and stock. Results under 200 ms, suggestions under 100 ms - Vehicle-first search: pick your car once, and every result says whether it fits - Part-number search in any format: dashes, dots, spaces, OE numbers and competitor cross-references - How people actually type: "06 malibu struts", "front brakes 2015 f150", typos, slang, synonyms - Ranking that sells: fits your car, in stock and priced first, then best sellers - No dead ends: zero results shows "did you mean", close part numbers, or "call us, we'll find it" - A nightly report of searches that found nothing or got no clicks, and fixing the top ones every week - Search analytics: zero-result rate, search to add-to-cart, and speed. Watched daily 2. Product discovery - Category, brand and vehicle pages ("parts for your 2015 F-150") for people who don't know the part name - Filters that make sense for auto parts (position, brand, kit vs single, price, in stock) - Related parts, "complete the job" kits, frequently bought together 3. Remarketing flows (technical side) - Clean tracking events (view, search, add to cart, checkout, purchase) to GA4, Google Ads, Meta and Klaviyo, server-side where it matters - Wiring the flows: abandoned cart and browse, back in stock, price drop, "your car is due for..." from the customer's saved vehicle, and win-back - Every email or ad that brings someone back lands on the right part for their car 4. Dashboards - One live funnel dashboard: traffic by source, search, product page, cart, checkout, order, on phone and desktop - Search health, speed and flow results in one place, so you can see in 10 seconds what broke or what's working 5. SEO and AI search (technical) - Fast, crawlable pages across millions of URLs: server rendering, Core Web Vitals, sitemaps, canonicals, internal links, indexing - Structured data: Product, Offer, vehicle fitment, FAQ and Review schema - Making our pages the answer in Google AI Overviews, ChatGPT, Perplexity and Gemini - Search Console and crawl monitoring, and fixing what Google flags YOU ARE - Using Claude Code (or similar agentic tools) every day as your main way of building, with careful review of AI output - Strong in Next.js / React / TypeScript, comfortable in Python (FastAPI) and MySQL - Hands-on with a search engine at scale (Typesense, Algolia, Elasticsearch): synonyms, typo tolerance, ranking, facets, keeping a big index in sync - Experienced with e-commerce tracking and flows: GA4, Tag Manager or server-side tagging, Klaviyo, Google and Meta conversion tracking - Strong in technical SEO for big sites, and you follow how AI search is changing it - Product-minded: you can explain why RockAuto, carparts.com or AutoZone converts, then build better - Data-driven: every change comes with before/after numbers - Bonus: auto parts, fitment data (ACES/PIES), or any catalog with millions of products FIRST 30 DAYS 1. Get our catalog into Typesense and switch search to it behind an on/off switch 2. Fix part-number search end to end 3. Ship the "no dead ends" results page and the nightly zero-results report 4. Audit tracking end to end (does every purchase reach GA4, Ads and Klaviyo?) and fix the gaps 5. Ship the first funnel dashboard Write "ClaudeMax" in the top of your proposal. DAYS 30-90: vehicle-first results everywhere, abandoned cart and "your car is due" flows live, product and fitment schema on every product page, and a monthly AI-search visibility check.