How Mamo's Engineering Teams Use AI

Written with love by

Imad Gharazeddine

At Mamo, we're not just building the future of Fintech; we're building it smarter. We constantly explore how we can leverage cutting-edge technology to accelerate our development cycles, enhance code quality, and ultimately, deliver a superior experience to our users. A key part of that strategy? Embracing AI-powered coding assistance.

We've seen firsthand how tools like Gemini Code Assist and GitHub Copilot can transform our engineering workflow. But simply throwing these tools at our developers isn't enough. We've taken a strategic approach. We empower our team with the knowledge and techniques to truly maximize their potential.

In this post, we wanted to share these techniques with you all, to help enhance your engineering teams and build cooler products faster!

Move Beyond Basic Code Completion

It's not just about generating boilerplate code anymore. At Mamo, we push the boundaries of what these AI assistants can do. When we craft precise, structured prompts, that becomes the key to unlock their true power.

We train our AI assistants to be our expert pair programmers. We guide them with clear, context-rich instructions, and ensure they understand the nuances of our Fintech domain.

Our Prompt Engineering Philosophy

We've adopted a structured approach to prompt creation, emphasizing clarity and specificity:

Context is King (Primer): To begin, we establish AI's role, emphasize its expertise in writing secure, efficient, and compliant Fintech code. For example, "You are an expert in writing secure and performant Python code within a Fintech environment."
Clear Objectives (Question): We precisely define the task at hand, whether it's generating a specific function, refactoring a module, or debugging a complex transaction flow.
Detailed Instructions (Decorator): We provide explicit instructions on the desired output, format, and process. We encourage our developers to ask the AI to "show me" its reasoning, step-by-step, rather than simply "give me" the code.

Real-World Applications at Mamo

Here are just a few examples of how we're leveraging AI assistance at Mamo:

Accelerated API Development: We use prompts to rapidly generate secure and efficient API endpoints, and adhere to our strict compliance standards.
Robust Test Case Generation: AI assists in creating comprehensive test suites, ensuring the reliability and security of our financial transactions. Of course, all test suites and unit tests are checked by a developer and code reviewed diligently.
Proactive Debugging and Optimization: We leverage AI to identify potential performance bottlenecks and security vulnerabilities. This helps us to address them proactively.
Document Generation and Refactoring: AI generates clear and concise documentation, and helps us refactor older code to meet our stringent standards.

Our Tooling and Integration Strategy

To make this seamless, we've implemented the following:

Customized Copilot Profiles: We've created tailored Copilot profiles with Mamo-specific prompts and configurations, ensuring consistency across our engineering team.
Repository-Specific Prompt Repositories: We have created repositories for common prompts, ensuring our specific needs are met.
Scripting Automation: We use scripting to help automate the use of complex prompts, and to integrate AI assistance into our CI/CD pipelines.

The Future of Fintech Engineering

It may appear obvious to most, but we’ll say it anyway; We believe that AI is not just a tool, it's a strategic advantage. By empowering our engineers with the ability to effectively collaborate with AI assistants, we speed up innovation, enhance code quality, and — ultimately — deliver a superior Fintech experience to our users.

‍