AI Engineer — LLMs, RAG and Engineering Agents

About the role

This role covers applied engineering AI at PhDsoft. The focus is to turn AI into real engineering-software functionality, with evidence retrieval, RAG, Graph-RAG, agents, a capability catalog, model evaluation, confidence criteria and human validation.

The person will work on applied engineering problems, not only abstract research. The aim is to create AI capabilities that help engineers query data, interpret documents, invoke WDT capabilities, assess answers, recognize limits and preserve technical responsibility.

Main responsibilities

• Develop AI pipelines applied to the WDT, including RAG, Graph-RAG, embeddings, vector databases and knowledge retrieval.
• Design agents, tool and function calling, use policies, context memory and an access-controlled capability catalog.
• Create evaluations, benchmarks, confidence criteria, abstention mechanisms and answer validation.
• Integrate AI with APIs, backend, technical documents, simulations, inspections, sensors and engineering workflows.
• Support parsing, OCR, cleanup of technical corpora, source versioning and provenance.
• Work with domain engineers to turn technical rules, risk criteria and evidence into usable capabilities.
• Document architecture, limitations, tests, results and acceptance criteria.

Required qualifications

• Practical experience with Python and with AI applications or pipelines.
• Knowledge of LLMs, model APIs, embeddings, RAG and answer evaluation.
• Ability to turn models into services, integrations or real features.
• Familiarity with machine learning, document processing, information retrieval and prompt engineering.
• Concern for traceability, security, human validation and model limitations.
• Ability to explain technical decisions to software and engineering teams.

Preferred qualifications

• Experience with Graph-RAG, knowledge graphs, agents, tool/function calling, MCP or vector databases.
• Knowledge of MLOps, serving, observability, benchmarks and A/B testing.
• Background in engineering, simulation, technical documents, FEA, BIM, geotechnics, naval work or asset management.
• Experience with computer vision, multimodality, OCR or inspection-image analysis.

Related technologies and tools

Python, LLM APIs, RAG, Graph-RAG, embeddings, vector databases, evaluations, agents, tool/function calling, OCR/parsing and APIs.

Profile

We are looking for a pragmatic, rigorous and curious person who can build AI that helps technical professionals without replacing human validation. They should like measuring quality, recognizing uncertainty and turning prototypes into product.

How to apply

To apply, send the following materials:

How do you currently use artificial-intelligence tools in your professional workflow? Name the tools, give examples of activities, and explain how you validate the results produced with AI assistance.

Additional information may include LinkedIn, GitHub, Behance, Google Scholar, articles, publications, websites, demonstrations, applications or other links relevant to the role.

• An up-to-date curriculum vitae with education, professional experience, main projects, technologies used and responsibilities held.
• A portfolio with concrete examples of previous work. For software, this may be GitHub, applications, systems, APIs, interfaces or projects that can be demonstrated. For engineering, this may be models, simulations, reports, analyses, publications or technical work that can be shared. Confidential projects may be presented in anonymized form.
• A presentation video of up to 5 minutes, briefly showing a project or portfolio, the problem solved, your contribution, the technologies used, your working workflow, everyday tools, your use of artificial intelligence and how you validate results produced with AI support.