OpenSearch Consultant - May 2026

Stellenbeschreibung:

For one of our clients in the fashion industry, we are looking for a freelance OpenSearch Consultant – May 2026 .

Overview

Creation of an in‑house solution for on‑site search along with a Merchandising UI.
The project aims to replace the current keyword‑driven external solution for on‑site search with an AI‑driven solution featuring semantic understanding and personalization capabilities.
The services mentioned in point 3 will be delivered within the framework of the agile development method Scrum.

Background

The company is developing a next‑generation, AI‑driven on‑site search and PLP‑ranking capability. Several critical elements require specialized expertise in:

OpenSearch relevance engineering
Multimodal product embeddings
Semantic search optimisation
Personalisation models
LLM‑based query processing
Hybrid lexical/semantic retrieval
Multilingual search infrastructure

This expertise is not available internally. The contractor therefore provides a unique contribution with responsibilities significantly different from internal staff.

Tasks

Technical consultation, configuration, and optimisation of OpenSearch‑based search relevance components, including analyzers, scoring parameters, hybrid retrieval structures, and vector‑search integrations.
Development and refinement of retrieval and ranking models, including multi‑stage ranking approaches, learning‑to‑rank concepts, and integration of relevance, behavioural, and business signals into ranking pipelines.
Creation of multilingual NLP components for various locales (LAM, EU, NAM), including tokenisation, stemming, normalisation, and locale‑specific linguistic processing within OpenSearch and related pipelines.
Design and implementation of query understanding mechanisms, including synonym and concept extraction based on catalog and interaction data, query intent interpretation methods, and term/concept expansion techniques.
Development of LLM‑enhanced search components, including prompt construction and incorporation of LLM‑derived semantic signals into retrieval and ranking logic.
Creation of personalisation logic for search and PLP ranking, including re‑ranking frameworks balancing user affinity, semantic relevance, and commercial parameters.
Development and validation of autocomplete, spelling correction, and search suggestion components, ensuring robust multilingual handling and adherence to domain‑specific terminology.
Definition and refinement of commercial relevance models, including recency‑weighted popularity signals, interaction‑based relevance indicators, and business‑driven ranking adjustments.
Construction of evaluation and diagnostic frameworks for relevance quality using offline IR metrics and analytical assessment methods.
Modelling and integration of real‑time or near‑real‑time data signals (e.g. stock levels, size availability) into filtering, faceting, and ranking components.
Technical implementation of data‑science‑driven backend logic for the Merchandising UI, including scoring routines, rule evaluation structures, configurable business logic, and interfaces for merchandising adjustments to search and ranking behaviour.