Abstract
This paper chronicles the collaboration between a senior computational chemist-spanning four decades of computing evolution from 1983 punch cards to 2025 artificial intelligence (AI)-assisted infrastructure-and a high-performance computing (HPC) expert, mediated by AI tools, culminating in the successful deployment of a multi-node research computing cluster for molecular modelling and drug design. An initial Gaussian 16 software request expanded into a comprehensive cluster implementation using Rocky Linux 9, SLURM workload management, parallel filesystems, and two key computational chemistry packages (Amber24 and Gaussian 16), with ORCA, GAMESS-US, Psi4, NWChem, and CP2K identified as suitable candidates for future installation; the cluster optimises mixed graphics processing unit (GPU) architectures (NVIDIA RTX A4000/RTX 4060), a common reality in resource-limited laboratories. A structured multi-AI collaborative methodology-utilising Claude AI for documentation, Grok for alternative perspectives, and DeepSeek for technical verification-resolved approximately 150 distinct technical challenges with an 85% first-attempt success rate, completing deployment in approximately 6 months compared to the estimated 12-18 months required by traditional approaches. Benchmarks confirmed 99%+ Amber24 test suite validation, a 20 & times; GPU speedup for large molecular systems (- 1 million atoms), and 17% GPU acceleration for quantum chemistry calculations. The AI approach thrived via an iterative model of problem-solving, explanation, and reasoning, with human-led execution, validation, and adaptation essential throughout. All Warewulf cluster installation manuals are provided as Supplementary Material (S1-S27), enabling any researcher to assemble a comparable system with modest effort and a proficient AI collaborator, thereby removing barriers to HPC adoption and heralding a paradigm shift in scientific knowledge transfer for resource-limited institutions.