GenAI Engineer – Chat with Data & NL2SQL

PKO BP Finat develops the "chat with data" solution enabling natural language question answering on data and the Unstructured Data Repository (UDR) platform for processing, storing, and sharing unstructured data such as documents, images, and audio recordings used by AI solutions.

• Very good knowledge of Java and/or Python
• Practical experience building solutions using large language models (LLM)
• Familiarity with at least one framework such as LangChain4j, Google ADK, Pydantic AI, or Spring AI
• Knowledge of Google BigQuery including SQL, data modeling, and optimization of costs and performance
• Experience creating services and REST API integrations
• Ability to test and diagnose LLM-based systems
• Understanding of best practices for software creation, deployment, and maintenance

Nice to have:

• Experience with NL2SQL, conversational analytics, or BI-as-chat solutions
• Knowledge of RAG, embeddings, chunking, and re-ranking
• Experience processing documents, images, or audio, including OCR, text extraction, and transcription
• Experience with GenAI system evaluation, offline and online tests, and prompt versioning
• Familiarity with semantic or metrics layers, caching, and pre-aggregation
• Knowledge of MCP and agent architectures
• Experience with vector databases like Qdrant, Pinecone, Weaviate, or Milvus
• Familiarity with graph databases such as Neo4j and GraphRAG approach
• Knowledge of FastAPI, Kubernetes, Cloud Run, and Pub/Sub
• Knowledge of Rust
• Experience in regulated environments involving audits, data classification, access control, and sensitive data protection

• Design and develop the "chat with data" solution, especially NL2SQL and conversational analytics mechanisms
• Build services utilizing large language models (LLM) and integrate models with company data and systems
• Create safeguards against hallucinations, erroneous queries, and unauthorized data access
• Develop Retrieval-Augmented Generation (RAG) solutions including chunking, embeddings, semantic search, and re-ranking
• Design tests and evaluate the quality of GenAI solutions
• Optimize quality, performance, and costs of BigQuery queries
• Develop services and API integrations in Java and/or Python
• Support UDR in processing documents, images, and audio recordings, extracting content and metadata, and sharing results with AI solutions
• Integrate AI models with unstructured data processing workflows
• Implement, monitor, and maintain solutions using DevSecOps practices

Data administrator is PKO BP Finat sp. z o.o., Warszawa, ul. Chmielna 89. Candidates have rights to data access, correction, deletion, processing restriction, objection, data portability, and complaint submission. Data is processed for recruitment purposes, mandatory as per Polish Labor Code. Consent for optional data is voluntary and can be withdrawn anytime. Data will be stored until recruitment ends and possible claims are resolved. Data recipients include Just Join IT and contracted processing entities.