Why Better Models Are Not Enough for AI Coding
Why reliable AI coding depends as much on clear module boundaries, executable rules, and CI as it does on capable models.
- AI Coding
- AI Agents
- Software Architecture
- Modularity
- Testing
- CI/CD
Passionate software engineer with 7+ years of experience building scalable web and mobile applications. Currently leading UI/UX and frontend development at Marex, specializing in React, TypeScript, and modern architectures. I thrive on solving complex problems and mentoring teams to deliver high-quality software.
Leading UI/UX and frontend development for the new Neon Payments product.
Led frontend development for trading platforms using React, TypeScript, and micro-frontend architecture. Mentored junior developers and drove technical decisions for a team of eight engineers.
Developed and maintained trading applications using React, .NET, and Azure. Implemented CI/CD pipelines and improved code quality through SonarQube integration.
Built full-stack features using C#, .NET Framework, and React. Collaborated with cross-functional teams to deliver customer-facing trading tools.
Worked on enterprise software solutions while completing my studies, gaining experience with Python, SQL, and agile delivery.
A selection of recent work I'm proud of, spanning mobile, web, and AI solutions.
On YouTube
On devBarbar I talk openly about work, personal growth, procrastination, health, and learning to move forward before everything feels perfectly figured out.
Visit the devBarbar channelBarbar Ahmad's blog
Honest reflections and practical ideas from work, technology, and life.
Why reliable AI coding depends as much on clear module boundaries, executable rules, and CI as it does on capable models.
Why even the selfie stick stayed in the car, what my reaction to a friend's video taught me, and why I am still going to keep recording.
Why professional success does not automatically mean that your life feels like it is moving forward, and why I am starting this channel anyway.