Python Developers to Support the Autonomous Resilient Cyber Defence (ARCD) Project

Description

Summary of the work Dstl requires a team of Python developers to contribute to the development of the Cyber Autonomy Gym for Experimentation (CAGE) and provide software engineering support to an internal project team evaluating open source reinforcement learning tools. Expected Contract Length 12 months (There is an option to extend the duration by 25%, although this is not guaranteed.) Latest start date Monday 3 January 2022 Budget Range £1,000,000.00 Why the Work is Being Done The Autonomous Resilient Cyber Defence (ARCD) project aims to develop self-defending, self-recovering concepts for military platforms and technologies. The project is seeking a team of Python developers to: contribute to the development of the Cyber Autonomy Gym for Experimentation (CAGE), a 5-eyes initiative under The Technical Corporation Program (TTCP); provide software engineering support to an internal Dstl project team reviewing and assessing emerging open source reinforcement learning tools. Problem to Be Solved The CAGE project aims to provide a platform for development and validation of autonomous, Artificial Intelligence (AI) driven agents for cyber defence. The scope of work on CAGE includes (but is not limited to): implementation of a state space visualisation capability for the CAGE simulator; expanding the scope of the Action Set for the CAGE simulator; and implementation of specific agents within the CAGE framework. The supplier will be required to document their progress and brief the Dstl project team (and international partners) on progress. Who Are the Users The scope of software engineering support to the Dstl project team will include (but is not limited to): identification, assessment and documentation of new and emerging reinforcement learning tools for cyber security; supporting the development of abstract and emulated cyber security reinforcement learning environments, which can be shared with Industry and Academia; and development of common libraries and tools to support reinforcement learning research activities. Existing Team The supplier will be working collaboratively with both an internal Dstl team and international partners based in Australia and the US West Coast. Personnel will be supervised by the ARCD Project Technical Authority (PTA) and Project Manager. Current Phase Alpha Skills & Experience • Demonstrable experience of developing robust, high quality software solutions in demanding timeframes, within the quoted budget, in the Python 3 language. • Demonstrable experience of third party technology integration. • Demonstrable experience of developing and implementing software test plans. • Demonstrable experience of authoring high quality software documentation (including commented code). Nice to Haves • Experience using Jupyter Notebook (or similar) is desirable. • Familiarity with OpenAI Gym is desirable. • Familiarity with the Atlassian software suite (particularly Jira and Bitbucket). Work Location "Dstl Porton Down Salisbury Wilts SP4 0JQ Whilst it may be possible to conduct aspects of the work remotely (using Dstl accredited ICT), it is essential that the supplier provides personnel that are able to work on site at Dstl Porton Down. Access to site will be managed in line with Dstl's COVID-19 guidance." Working Arrangments Work conducted collaboratively with Dstl staff will be managed in an Agile manner using the Kanban approach. The supplier will work with the Dstl internal team to develop a prioritised project backlog that will be hosted on Dstl’s internal Jira instance. Work conducted with international partners will utilise the DI2E platform hosted by partner organisations in the US. The DI2E platform provides a central code repository. Developers from contributing organisations develop features locally before pushing back to the central repository. All work must be conducted on Dstl accredited ICT infrastructure at appropriate classifications. Security Clearance Minimum of SC clearance. Personnel must be British Nationals. Personnel with dual nationalities will not be considered. No. of Suppliers to Evaluate 5 Proposal Criteria • Demonstrate with evidence the ability to quickly familiarise with a complex inherited codebase. (0.2) • Demonstrate expertise and experience of developing software to implement machine learning and AI algorithms. (0.2) • Demonstrate expertise and experience of evaluating and testing machine learning and AI systems and tools. (0.2) • Demonstrate experience of developing software in the context of the cyber security domain. (0.2) • Provide a breakdown of the team structure (please include CVs of the team members doing the work). (0.2) Cultural Fit Criteria • Demonstrate with evidence ability to follow industry best practice throughout the whole software development lifecycle, from requirements gathering through to documentation, testing, verification and validation (including examples of tool-chains used).(0.2) • Demonstrate consistent cultural commitment to agile software development practices. (0.2) • Demonstrable experience of working-collaboratively with external-organisations to realise project-goals, and responding to evolving-requirements. Show evidence of being transparent and collaborative both internally and with the customer when making decisions.(0.2) • Show evidence of an internal culture of knowledge and experience sharing. (0.2) • Follow industry best practice when conducting Verification and Validation for example TickITplus. (0.1) • Demonstrate ability to successfully deliver within the UK government customers. (0.1) Payment Approach Capped time and materials Evaluation Weighting Technical competence 60% Cultural fit 20% Price 20% Questions from Suppliers 1. How many Python R resources are estimated to be required initially and will they be required on full time basis?Will DSTL hold the required security clearances? Dstl is looking for suppliers to put forward a team of approximately 4 Full Time Equivalent (FTE) staff for the 12 month duration of the task. There is no requirement for any individual member of the team to work full time on the task. For clarity, Dstl is looking for Python developers – experience with R is not required.Dstl expects proposed personnel to already hold security clearances or for them to have them prior to commencing support of the task.