Strategic Search & AI Architect

Marktplaats; Amsterdam; On-site; EUR 120,000 - 180,000; Full time

Job summary

Marktplaats, part of Adevinta, is seeking an architecture leader to own the search and AI stack for our Benelux marketplaces. You will define multi-year visions, drive semantic search, and shape GenAI usage at consumer scale.

Amsterdam-based role with strong cross-functional collaboration, bridging data and product teams, and delivering measurable improvements while maintaining governance and quality.

Qualifications

• 10+ years in software engineering or ML, with 5+ years designing search, relevance or ML systems at consumer scale.
• Deep expertise in search and relevance architecture at consumer scale, spanning retrieval, ranking, embeddings, and evaluation.
• Practical GenAI experience beyond demonstration, with production pattern knowledge.

Responsibilities

• Own the search, matching, and relevance architecture: target state and evolution path.
• Define multi-year vision for search including LLM-based intent modelling and retrieval.
• Establish guardrails for GenAI in product experiences: cost, risks, and governance.
• Represent consumer needs to ML/AI platform in partnership with data teams.
• Translate ML/AI requirements into data quality, lineage and governance expectations.
• Navigate multiple models and providers, deciding standardisation vs plurality.

Skills

Search architecture

ML systems

GenAI experience

Computer vision

Delivery by influence

Cross-functional teamwork

Tools

Databricks

ElasticSearch

Docker

Kubernetes

Scala

Go

Java

Job description

About Marktplaats

Marktplaats is one of the Netherlands' leading online marketplaces, helping make commerce more sustainable by giving products an extended life. With over 2.1 million buyers and sellers visiting our platform every day, around 350,000 new listings added daily and users spending an average of 22 minutes per visit, we’re building a platform that people use and trust. Every connection we enable helps people save money, earn money and reduce waste. As part of Adevinta, one of Europe's leading online classifieds groups, we’re building products that make buying and selling second‑hand products simpler, safer and better for everyone.

This role sits within our Product & Technology function in our Amsterdam office, home to over 300 talented professionals, where cross‑functional teams design, develop and scale the products that millions of users rely on every day across our Benelux marketplaces, Marktplaats and 2dehands/2ememain. You’ll join a small architecture team, reporting to the Head of Architecture, embedded with the search and AI product teams and working closely with the data organisation. The foundations are strong: ADRs and an experimentation culture are already in place. The next chapter is a generational shift from query‑matching to searcher‑understanding, and on a marketplace, that means words and images in equal measure: visual search, image understanding of listings, and camera‑based discovery. This shift needs an architectural owner.

What You Will Do

• Own the search, matching, and relevance architecture: the target state, the key design decisions, and the evolution path from today's stack to one that carries semantic and AI‑driven experiences at consumer scale.
• Define the multi‑year vision for the future of search, spanning LLM‑based intent modelling, conversational and agentic retrieval, persistent user memory, behavioural signal fusion in ranking, and computer vision in the experience. Determine which research threads matter for a marketplace, and sequence the ones that do.
• Establish patterns and guardrails for GenAI in the product experience, covering retrieval, ranking, and conversational interfaces: what good looks like, what it costs, and when not to use it. Build on the existing ADR base.
• Represent the consumer side of the ML and AI platform in partnership with the data organisation, articulating what product experiences need from the platform.
• Translate search and AI requirements into clear, actionable expectations around data quality, lineage, and governance. Advise and support the data organisation where asked, without owning their platform.
• Bring clarity to a deliberately varied AI landscape: models served from Databricks, LLMs from several providers. Deciding what we standardise on, and what stays plural, is part of the job.

Who You Are

• At least 10 years of experience in software engineering or applied ML, including 5+ years designing search, relevance or ML systems at consumer scale.
• Deep expertise in search and relevance architecture at consumer scale, spanning retrieval, ranking, embeddings, and evaluation.
• Able to track the research frontier and translate it into product reality: conversational query rewriting, agentic multi‑step search, persistent user memory, LLM‑based intent modelling, and unbiased learning‑to‑rank. Knows which of these ship and which stay papers.
• Practical GenAI experience beyond the demo stage, with a clear grasp of the patterns, costs, and failure modes that only emerge in production.
• Experienced with computer vision, including visual search, image similarity, and image understanding. A strong plus for this role: a marketplace is a visual product.
• Proven ability to shape ML systems in close collaboration with data teams without needing to own their platform, maintaining a healthy working relationship across an organisational boundary.
• Designs with evaluation built in: offline metrics, online experiments, LLM‑based simulation where live traffic is too slow, and the discipline to kill features that do not perform.
• Pragmatic, delivery‑credible, and effective through influence, given that most of the teams you will shape do not report to you.

Bonus Points

• Experience with marketplace or e‑commerce search, where inventory is unique, short‑lived and cold‑start is constant
• Vector search and embedding infrastructure at scale
• Multimodal retrieval that combines text and images
• Unbiased learning‑to‑rank and click‑model debiasing
• Experimentation platforms and offline evaluation harnesses
• Contributions to open‑source search or IR tooling, or publications and talks in the field
• Dutch is not required; we work in English

Our Tech Stack

• Languages: Scala, Go and Java on the back end; Kotlin, Swift and TypeScript on the client side
• Databases: MySQL on Aurora RDS and Cassandra, among others
• Search and AI: ElasticSearch; models served from Databricks (Mosaic AI); LLMs from Gemini, OpenAI and AWS, via Bedrock, OpenAI and Databricks
• Data platform: Databricks
• Cloud: AWS
• Other tools: Docker, Kubernetes, Terraform, Kafka, RabbitMQ, GitHub, DataDog

This role is based out of our Amsterdam office. While we offer flexibility, we value in‑person collaboration and encourage our teams to spend regular time in the office.

Our DNA

Ownership: We take ownership, step in where needed and make things happen.

Engagement: We show up for each other, build meaningful connections, celebrate success and enjoy the ride together.

Customer Focus: We put our customers at the heart of our decisions, back our ideas with data and deliver value that makes a difference.

Execution: We move with purpose, challenge ideas respectfully, act with confidence and deliver at pace.

Curiosity: We ask questions, challenge the status quo, embrace learning and turn insights into better decisions.

Equal Opportunity Statement

Marktplaats is an equal opportunity employer, and we value diversity. We do not discriminate on the basis of race, religion, colour, national origin, gender, sexual orientation, age, marital status, or disability status.