Software Engineer – AI Applications
CONCELEX · Bucharest Metropolitan Area
About this role
The candidate will own AI applications end-to-end, from concept to production, focusing on building retrieval engines, document processing pipelines, and business-wide AI applications. The role involves engineering prompts, retrieval, and agent workflows with a heavy emphasis on evaluation. The technical environment includes building backend services in Python and internal frontends using React or similar, working closely with Data Engineers, Cloud Platform Engineers, and Product Managers. The engineer will operate within an EU-hosted model endpoint environment to ensure data security and zero data retention. The position requires a product mindset to ensure high adoption and user value.
Skills & technologies
Must have
- Python
- RAG
- Embeddings
- Retrieval
- Prompt Engineering
- LLM Evaluation
- React
- Full-stack
Nice to have
- Agentic AI
- MCP
- Agent Frameworks
- OCR
- Azure OpenAI
- Azure AI Foundry
- Databricks Model Serving
- SQL
- dbt
Read full description
Purpose of the role
Own the company's AI applications end-to-end.
You will build the products this transformation programme is ultimately about: retrieval engines across years of company documents, document processing pipelines that turn files into structured data, and AI applications used across the business every day.
Working within the Data & AI platform and architecture, you will own what users touch, what they adopt and the value they get from it.
Responsibilities
- Own AI applications from concept to production, ensuring adoption, quality and business impact.
- Sit with your users weekly: watch how they actually work, ship improvements and measure whether they are used.
- Engineer prompts, retrieval and agent workflows with evaluation as a habit, not an afterthought.
- Build backend services in Python and a clean internal front end (React or similar), without a designer.
- Consume the Gold layer and the document store the Data Engineer maintains, and feed requirements back.
- Run everything on approved EU-hosted model endpoints configured for zero data retention: commercially sensitive data never leaves the perimeter.
- Keep token and compute costs proportionate to the value shipped.
- Work daily with the Data Engineer on retrieval-ready data, the Cloud Platform Engineer on deployment and security, and the Product Manager on what users actually need. Document as you go.
Qualifications & Experience
- 5 to 7 years of software engineering experience, including designing, building and operating production services in Python.
- Experience delivering at least one LLM-powered product or feature to real users, with hands-on knowledge of RAG, embeddings, retrieval, prompt engineering and evaluation.
- Full-stack capability, with experience building internal applications and user-facing interfaces using React or similar frameworks.
- Strong understanding of LLM evaluation, including quality measurement, test set creation and failure analysis.
- Product mindset and experience working directly with business users, turning real problems into adopted solutions.
- Ability to balance speed, quality and pragmatism when developing AI products.
- An AI-first way of working, or strong enthusiasm to build one, using modern AI development and agent tools responsibly and effectively.
Nice to have
- Experience with agentic AI systems, including tool use, function calling, MCP, agent frameworks and evaluation frameworks.
- Experience with document processing, OCR and extracting structured information from complex documents.
- Experience with Azure OpenAI, Azure AI Foundry, Databricks Model Serving or similar AI platforms.
- Exposure to construction, engineering or other document-intensive industries.
- Working knowledge of data engineering concepts, including SQL and dbt.
