Data-Centre-Grade AI-HPC Server
Dell PowerEdge · Intel Xeon Gold · NVIDIA L40S
Hardware Configuration
🗄️Dell PowerEdge Serverdata-centre-grade chassis
🧠Intel Xeon Gold64 cores · server-class CPU
🎮NVIDIA L40S GPU~13,000 CUDA · ~700 Tensor cores · 48 GB VRAM★ STAR
🧬NUMA Architectureoptimised for HPC computing
⚡Primary StorageNVMe SSD · high-speed
💾Secondary StorageSATA SSD · bulk capacity
🌐Network Interface10 Gigabit Ethernet
What students & departments can build on it
One server, many possibilities — grouped by what they're for.
🧠
LLMs & Generative AI
🦙Locally-hosted open-source LLMs — Llama 3, Mistral for private on-prem reasoning.
⚡vLLM inference server — high-throughput serving of open LLMs.
🕸️Advanced AI systems — Multi-Agent Systems, RAG, MCP, LangGraph.
🔬
AI Application Development
📊Classical ML — predictive models, regression, classification.
👁️Computer Vision — detection, segmentation, image analysis.
🧪AI app platform — build & ship end-to-end AI applications.
🏗️
Model Training & MLOps
🚀Large-scale training pipelines — multi-GPU data pipelines.
⚙️Training optimization — mixed precision, DataLoader tuning, profiling.
🚀
Platform, DevOps & Hosting
🐳Dockerised application development
☸️Minikube (Kubernetes) deployment
📓JupyterHub — browser-based notebooks for everyone.
🌍Web hosting — college apps & student projects.
🎓
Multi-Department Access
🏫Shared across all engineering streams — Mechanical, Chemical, Electrical & more; each builds domain-specific AI.
⚡Example — Electrical Dept: Electrical Transformer Failure Prediction.
⚙️ Mechanical🧪 Chemical⚡ Electrical🤖 Robotics🏗️ Civil
🛡️
Operations, Security & the AI Ecosystem
📈Server observability — live CPU / GPU / memory / storage usage.
🔒Network security monitoring & hands-on cybersecurity practice.
🌐Production-Grade AI Ecosystem platform — AI + Linux + technology domains (Edge/Cloud Computing, Big Data, Cybersecurity, Distributed Systems).