Designing and building a custom life insurance projection and asset and liability management (ALM) model in-house is a challenging endeavour that many insurance companies are considering. This white paper investigates numerous challenges and decisions that must be taken regarding the architectural choices that fit different functional requirements, as well as implementation approaches that ensure high performance and code simplicity. In the paper, we discuss: 

• The temptation of an in-house model and risks involved
• Architectural and functional choices in cash flow projection models 
• Information technology landscape
• Implementation in Python
• Our benchmark