Back to SearchSenior Python Backend Engineer (LLM/RAG)
Hybrid in Singapore
Python.AI& 5 others
Looking for something else?
Find a vacancy that works for you. Send us your CV to receive a personalized offer.
Find me a job
We are seeking a Senior Python Backend Engineer (LLM/RAG). You will build APIs and event-driven services that power AML and sanctions screening, combining NLP, search and vector retrieval to improve match quality and auditability. Work closely with product and compliance partners in a hybrid setup in Singapore.
Responsibilities
- Build and maintain Python back-end services, APIs and microservices for screening workflows
- Design event-driven integrations to support high-throughput, traceable processing
- Implement LLM and retrieval-augmented generation (RAG) patterns for entity resolution and investigation support
- Develop search and retrieval using PostgreSQL, Elasticsearch/OpenSearch and vector stores
- Partner with compliance and engineering stakeholders to translate controls into system behavior
- Improve observability, audit trails and explainability for screening decisions and overrides
- Write automated tests and support reliable releases in a regulated environment
Requirements
- Python proficiency for production back-end development
- Back-end architecture experience with microservices, APIs and event-driven systems
- Search and retrieval capability across full-text search and relevance tuning
- Databases knowledge with PostgreSQL and data modeling for traceability
- Vector databases familiarity such as Pinecone or similar vector stores
- Generative AI hands-on experience with LLM applications, RAG, agents or workflow-based orchestration
- AML screening working knowledge across KYC, sanctions, PEP, watchlists, adverse media and negative news
- Risk and controls mindset with focus on auditability, logging and explainable outcomes
Nice to have
- Experience with Azure AI Search
- Familiarity with OpenSearch tuning and relevance evaluation approaches
- Exposure to screening model governance, thresholding and investigation workflows