backend developer in semantic search

NDA; United States; Remote; USD 120,000 - 190,000; Full time

Job summary

NDA ищет опытного инженера поиска для контроля качества поисковых функций на крупном сайте. Работа включает управление как фильтрами, так и AI-поиском, настройку релевантности и показателей использования, а также развитие гибридного поиска на основе embeddings.

Вы будете работать в AI- natives команде с прямым взаимодействием с продуктом и руководством, реализовывать семантический и гибридный поиск, анализировать данные и метрики, а также внедрять новые подходы для масштабирования решений.

Qualifications

• 4+ года опыта в поиске, ранжировании або рекомендациях
• Умение писать на SQL и Python
• Опыт работы с Elasticsearch/OpenSearch в продакшн
• Опыт работы с embeddings, векторными индексами, гибридным ранжированием
• Опыт разработки на TypeScript/Node и backend
• Опыт использования LLM функций и оценки
• Опыт решения задач сопоставления сущностей и отраслевых таксономий
• Будет плюсом: опыт на маркетплейсе, e-commerce, или B2B SaaS поиске

Responsibilities

• Own search quality across filter search and AI search
• Build relevance benchmarks, labeled query sets, offline evaluations, and online metrics to track whether searches lead to saved results, refined searches, or abandonment
• Tune Elasticsearch mappings, analyzers, keyword and industry matching, scoring, synonyms, and handling of messy company names and domains
• Build semantic and hybrid search by embedding company and person profiles, running vector search alongside keyword search, and tuning how the results are blended
• Use embeddings to classify companies by industry, business type, and B2B/B2C, group similar companies, and power “companies like this” features
• Improve AI search through prompts, schemas, model selection, and evaluations
• Identify missing or incorrect data, measure its impact, and recommend what to fix, enrich, or buy
• Benchmark data coverage, accuracy, and freshness against external sources and turn the results into a roadmap
• Ship code across a TypeScript backend and Python analysis tooling
• Use AI to label data, analyze queries, and run evaluations at a scale that would be impractical to manage manually
• End-to-end ownership of search quality for a product used at large scale
• An AI-native team where shipping with agents is the default
• Direct work with product and leadership

Skills

SQL

Python

TypeScript

AI agents

embeddings

Elasticsearch/OpenSearch

Tools

Elasticsearch/OpenSearch

Solr

Vespa

Job description

Описание

Описания нет

Задачи

• Own search quality across filter search and AI search
• Build relevance benchmarks, labeled query sets, offline evaluations, and online metrics to track whether searches lead to saved results, refined searches, or abandonment
• Tune Elasticsearch mappings, analyzers, keyword and industry matching, scoring, synonyms, and handling of messy company names and domains
• Build semantic and hybrid search by embedding company and person profiles, running vector search alongside keyword search, and tuning how the results are blended
• Use embeddings to classify companies by industry, business type, and B2B/B2C, group similar companies, and power “companies like this” features
• Improve AI search through prompts, schemas, model selection, and evaluations
• Identify missing or incorrect data, measure its impact, and recommend what to fix, enrich, or buy
• Benchmark data coverage, accuracy, and freshness against external sources and turn the results into a roadmap
• Ship code across a TypeScript backend and Python analysis tooling
• Use AI to label data, analyze queries, and run evaluations at a scale that would be impractical to manage manually

Требования

• Build with AI coding agents such as Claude Code or Cursor every day, and use agents for analysis and labeling
• Have a product mindset: understand user intent and search failures, choose fixes that improve outcomes, and explain trade-offs to product and leadership
• Have 4+ years of experience in search, ranking, or recommendations
• Have production Elasticsearch or OpenSearch experience; Solr or Vespa also count
• Have production experience with embeddings, including model selection and evaluation, vector indexes, hybrid ranking, and managing index size, latency, and re-embedding costs at hundreds of millions of rows
• Have built relevance evaluations using judgment sets, precision@k, or nDCG, and changed a product based on their findings
• Have strong SQL and Python skills
• Be comfortable shipping in a TypeScript/Node backend with AI agents
• Have shipped LLM features, including prompting, structured outputs, evaluation harnesses, and cost and latency trade-offs
• Be comfortable with messy entity data, including company matching, deduplication, classification, and industry taxonomies
• Будет плюсом: experience on a marketplace, e-commerce, or B2B SaaS search team, entity resolution or learning-to-rank experience, experience evaluating or integrating third-party data vendors

Условия

• End-to-end ownership of search quality for a product used at large scale
• An AI-native team where shipping with agents is the default
• Direct work with product and leadership