Data Scientist, Retrieval & Agentic Systems — Hybrid Madrid

Kharon; Madrid; Hybrid; EUR 949,000 - 1,060,000; Full time

Benefits offered by this job

Discretionary annual bonus

Job summary

Kharon, based in Madrid, is seeking a full-time Data Scientist, Retrieval & Agentic Systems. The role focuses on owning Corestream, enhancing semantic search, and integrating risk insights into retrieval systems. You will co-lead narrative generation and roadmap delivery with Product and Engineering.

You will work in a hybrid setup (3 days on-site) in Madrid, applying Python, SQL, and graph-aware techniques to build robust risk intelligence solutions for clients.

Qualifications

• 3+ years of experience in Data Science or AI Engineering.
• Hands-on experience with semantic search techniques (BM25, vector search, reranking).
• Experience with RAG pipelines and agentic systems in production.
• Interest or experience in global security, financial crimes, sanctions, or related domains.

Responsibilities

• Own and evolve Corestream, Kharon's risk intelligence feed for customers.
• Redesign and improve semantic search capabilities for content discovery and retrieval.
• Integrate risk analysis outputs into the search layer to surface graph-based insights.
• Drive narrative generation pipeline from graph and analytics to written intelligence for customers.
• Collaborate with Product and Engineering to own the Corestream roadmap and advance it.
• Tune metrics to define search quality and narrative relevance.

Skills

Semantic search

BM25

Vector search

RAG pipelines

Agentic systems

Python

SQL

Knowledge graphs

Product alignment

Job description

Kharon, based in Madrid, is seeking a full-time Data Scientist, Retrieval & Agentic Systems. The role focuses on owning Corestream, enhancing semantic search, and integrating risk insights into retrieval systems. You will co-lead narrative generation and roadmap delivery with Product and Engineering.

You will work in a hybrid setup (3 days on-site) in Madrid, applying Python, SQL, and graph-aware techniques to build robust risk intelligence solutions for clients.