Responsibilities You will be working closely with our senior quantitative analyst to optimise the existing code for our internal and client-facing dashboards. You will: Develop our internal and client-facing dashboards (backend and frontend) Ensure production reliability and data integrity Document logic clearly and maintain version discipline Contribute to architecture decisions and system scalability Clean and optimise code, debugging Research and evaluate LLM models most suitable for our purposes (Claude, GPT 5, Gemini etc) Integrate a persistent GPT chat thread with long-term memory for receiving and learning from human feedback. This feedback will be used to fine-tune output consistently. Integration of backend dashboard with frontend Umbraco dashboard What We’re Looking For Must-have: At least 5 years of Python programming experience (mid-level at least) Experience with AWS services and Git Programmatic AI skills including Langchain, LangSmith Good software design skills for building with scalability in mind Proven track record of Umbraco projects Nice to have: Advanced AI Engineering skills for building even more customised AI workflows Nextjs + React + Typescript OR Django for frontend dashboard development Strong quantitative and programming skills (Python, R, C++, or similar), with experience in building trading models and systematic strategies. Contract This job demands full-time attention from the successful developer. The project is expected to last 2 months. Developer must be available for questions and meetings during Singapore work hours. How to apply Send us your CV and portfolio that illustrates at least all ‘must-have’ skills required for this role. Also state your expected monthly compensation range in EUR Please share a sample from a LangChain project you’ve built. Ideally something that shows: – Chain/agent structure – Prompt design and organisation – State or memory handling – Error handling and retries – Overall project architecture A GitHub repository or extracted core files are fine. If the original work is confidential, please share a simplified or anonymised version.