Description
Do you want to teach a bank’s document engine to read?
Every mortgage application comes with a stack of documents: payslips, employer statements, valuation reports, bank statements and IDs. Our extraction engine automatically turns these documents into structured data that can be used throughout the lending process.
Making that engine smarter is one of Ohpen’s key objectives for 2026, as we move from classical extraction rules towards AI/LLM-based extraction.
As a Data Extraction Specialist – AI, you’ll help drive that transition end-to-end. You’ll configure and optimise the engine, build the LLM-based extraction that replaces existing rules, benchmark the results to prove extraction rate.Briefing
Ohpen launched the world’s first cloud-native core banking platform. Within our Lending Suite, several components work together to process every document that comes in with a mortgage application:- DocStreet is our document processing platform. It receives, classifies and routes incoming documents.
- DTA2 is the data extraction engine within DocStreet. It reads each document and turns it into structured fields, such as income, employer and property value, which our clients’ lending processes rely on.
- Orbis is the engine that executes automated validations on the extracted documents.
WHAT YOU WILL DO
- Improve extraction rates. Configure and tune DTA2 across different document types and fields, prioritising those that matter most to our clients.
- Roll out AI/LLM-based extraction. Design, build, evaluate and deploy LLM-based extraction within the engine. This includes everything from prompt design and model selection to guardrails, evaluation sets and production rollout.
- Prove that it works. Build and maintain the benchmarking loop that shows where extraction has improved, where it has regressed and why. No improvement ships without evidence.
- Improve the engine itself. Identify and implement engineering improvements across DocStreet, DTA2 and Orbis.
- Work closely with the people who use it. Collaborate with internal stakeholders across Product, Engineering and Lending Operations, as well as directly with our clients.
About You
- You are analytically strong. You pick up new domains quickly and are comfortable being the person who figures out how things actually work.
- You have a technical engineering background, with a degree in Computer Science, Software Engineering, Data Engineering or a comparable field, and 1–5 years of hands-on experience.
- You are AI-native. LLMs are a normal part of how you work. You know how to design and iterate on prompts and how to build benchmarks that show whether a change actually helped.
- You are a proactive self-starter. In a small team, there is no one to wait for. You spot what needs doing, propose it and take ownership of getting it done.
- You are a strong communicator. You can explain technical trade-offs to clients, respectfully challenge colleagues when needed and keep stakeholders informed without being asked.
NICE TO HAVE
- Experience with Java and AWS, as our engine runs on them.
- Dutch language skills, as most of the documents we extract are in Dutch.
Familiarity with the Dutch mortgage market, ideally through previous experience working with Dutch mortgage products, processes or systems.
Why us?
- Ownership of a strategic 2026 objective from day one, in a team small enough that your work is visible to both clients and leadership.
- The opportunity to move a production system used by regulated lenders from rule-based to AI-based extraction.
- A role at the intersection of engineering, AI and the Dutch mortgage domain.
- Hybrid working from our office in the heart of Amsterdam.
- Competitive compensation and benefits.