The project is aimed at creating an AI agent for an existing Python BDD framework in a corporate banking environment. The solution is intended for automatic conversion of functional test scenarios into ready-made automated scenarios in Gherkin, using the existing codebase and intelligent mechanisms for reusing steps.
Our goal is to create an applied AI system that will accelerate the development of BDD auto-tests, reduce the cost of regression testing, and improve the standardization of test automation. We want to build a next-generation internal tool for testing and development teams that will combine the capabilities of LLM, RAG, ReAct with the bank's existing landscape, and enhance the synergy of BDD with AIPDLC and specification-based development.
Key Directions:
- Creation of an AI agent for generating Gherkin feature files based on functional scenarios
- Building a retrieval pipeline based on RAG and PostgreSQL + pgvector for searching relevant steps, templates, and reusable elements
- Integration with the existing Python BDD framework and step library for maximum reuse of the current codebase
- Development of a user interface for the solution, with a priority on an IDE plugin for convenient work by test automation engineers
- Formation of quality control mechanisms for generation, scenario standardization, and reduction of skips when using existing steps
- Creation of a technological foundation for further development of AI approaches in testing, engineering practices, and development lifecycle management
Technology Stack:
- Python 3
- FastAPI, Streamlit
- PostgreSQL, pgvector
- Apache Kafka
- Docker
- Linux
- HTTP, REST, GraphQL
- LLM, RAG, ReAct, embeddings, vector search
Requirements
- Practical experience in building RAG systems and working with vector databases, preferably PostgreSQL + pgvector
- Understanding of OOP principles, SOLID, and design patterns
- Understanding of basic principles of multithreaded and asynchronous programming, including GIL and asyncio
- Practical knowledge and skills in working with LLMs
- Experience with application containerization tools, including Docker
- Leadership qualities: ability to make decisions and take responsibility
Additional Requirements
- Understanding of BDD, Gherkin, test automation, and principles of step definition reuse
- Understanding of client-side web development principles
- Experience in developing IDE plugins or developer tools
- Experience in evaluating the quality of AI solutions: accuracy, retrieval quality, traceability
- Understanding of spec-driven development and the application of AI in SDLC / QA processes
Responsibilities
- Design and approve the architecture of the AI agent, applications, and integrations
- Develop and enhance backend components of the AI agent, shared libraries, and integration interactions
- Implement solutions based on LLM, ReAct, and RAG, including searching and selecting relevant "steps" from the existing library
- Integrate the system with internal services, repositories, knowledge bases, and corporate infrastructure
- Formulate requirements for building CI/CD, monitoring, logging, tracing, and quality control for the AI agent
- Conduct code reviews, maintain code standards, and participate in team communications
- Interact with automated testing specialists, QA engineers, and developers to clarify requirements and use cases
- Participate in the development of the solution's user interface in the form of an IDE plugin
Conditions
- A team of professionals ready to support your initiatives
- Opportunity to work with new technologies, including AI
- Competitive salary
- Corporate training and certification in technical and business areas
- Corporate entertainment programs
- Possibility of remote work within the Russian Federation
We await your resumes at recruiter contact
