Mr Vikram Bahl1
1Google Cloud, Singapore
Biography:
Vikram Bahl serves as a Principal Architect for HPC and AI Infrastructure at Google Cloud, driven by a passion for building impactful software and supporting APAC's research and public institutions. Their career has been marked by successfully architecting highly distributed and intricate backend systems, a skill set now pivotal in guiding clients through their research cloud journey. Vikram excels in translating complex requirements into effective cloud-native solutions, particularly in the realms of HPC, data engineering, applied machine learning, and strategic system architecture, ultimately enabling organizations to achieve their ambitious goals with Google Cloud's powerful capabilities.
Abstract:
Background
AI is revolutionizing how we approach research, opening doors to insights we never thought possible. However, many researchers struggle with complex coding challenges, juggling tools, and struggling to manage workflows. The technical barriers are overwhelming, especially when you're focussing on the science itself. This presentation cuts through the complexity to show researchers practical ways to develop, train, and deploy AI/ML models more effectively.
Method
Walkthrough of proven coding practices that make AI/ML work manageable. Think modular code that's easy to understand and modify, version control systems that ensure your work is reproducible, and the latest and greatest tooling available to researchers, from initial data exploration to model training & deployment. The key is learning how to choose and combine tools that actually make your work easier, not harder, while handling the demanding computational requirements that AI research involves.
Results
When researchers adopt these structured approaches and smart tool choices, the transformation is remarkable. Projects move faster, results become more reliable and reproducible, scaling experiments becomes straightforward rather than stressful. Most importantly, researchers can spend more time on the creative, scientific aspects of work rather than wrestling with technical roadblocks.
Conclusion
Success in AI driven research isn't just about having the best algorithms but it's about having the right methodologies and tools to bring those algorithms to life. By embracing structured coding practices and making informed choices about the technical toolkit, researchers can focus on what they do best: pushing the boundaries of knowledge and making discoveries that matter.