NIET AI Centre of Excellence
Welcome to NIET's fully operational AI learning platform.
We've created a complete cloud-based environment where students can train and use artificial intelligence models with powerful GPU computing.
What We've Built for You
Think of this as a computer lab in the cloud where each student gets their own workspace with powerful NVIDIA H100 GPU access. Our platform combines cutting-edge infrastructure with user-friendly tools, making AI model training accessible to everyone.
Cloud Infrastructure
Kubernetes-based deployment with enterprise-grade reliability and scalability
GPU Power
NVIDIA H100 GPU providing cutting-edge computing capabilities for AI workloads
Personal Workspaces
Isolated environments accessible from any browser, pre-configured and ready to use
Five Powerful Features at Your Fingertips
Personal Notebook Workspace
Jupyter Notebook environment accessible from any browser, like having your own computer lab available 24/7
AI Model Training
Fine-tune large language models like Mistral-7B on custom data using Lora/QLora techniques
Experiment Tracking
MLflow automatically records all training experiments, parameters, and performance metrics
Model Storage
Minio provides safe storage for trained models and datasets, accessible anytime
Fast Inference
Pre-trained models ready for instant predictions and real-time responses
Get Started in 3 Simple Steps
Access your AI workspace in minutes with our streamlined onboarding process. No complex setup required—just follow these steps and you'll be training AI models today.
01
Open the Login Page
Navigate to https://niet.thqindia.com/ in your web browser. You'll see the platform login screen.
02
Create Your Account
Click "Sign Up" and enter your student ID (student001, student002, etc.). Set a secure password and submit your registration.
03
Login and Start Working
Wait for admin approval (typically within 24 hours). Once approved, login to access your personal notebook workspace with GPU computing power.
Train Your AI Models
Complete Training Workflow
Students can use provided Jupyter notebooks to load pre-trained AI models like Mistral-7B, a state-of-the-art language model. Fine-tune them with custom data to adapt the model to specific use cases.
All training experiments are tracked automatically, and trained models are saved for later use. A typical model fine-tuning takes just 5-15 minutes on our GPU systems.
- Load pre-trained models instantly
- Customize with your own data
- Track performance metrics
- Save and version models
Fast AI Inference
Use trained models to generate text, answers, and predictions in seconds. Our optimized inference system delivers production-ready performance for real-time applications.
Quick Response Time
Get AI-generated responses in just 2-3 seconds, enabling real-time interactions and rapid prototyping
Test and Iterate
Experiment with different prompts and scenarios to optimize your AI applications
Application Integration
Use AI models in your own applications through our REST API interface
from openai import OpenAI
client = OpenAI(
base_url="http://10.10.0.10:8000/v1",
api_key="class-secret-key-2026"
)
response = client.chat.completions.create(
model="mistralai/Mistral-7B-Instruct-v0.3",
messages=[{"role": "user", "content": "Your question here"}]
)
print(response.choices[0].message.content)How the System Works
GPU Computing
Powerful NVIDIA H100 GPU shared among students for AI model training in minutes
JupyterHub
Personal isolated workspace with pre-installed AI libraries, accessible via browser
Model Training
Fine-tune Mistral-7B with custom data in 5-15 minutes, all automatically recorded
MLflow Tracking
Automatic lab notebook recording experiments, metrics, and model versions
MinIO Storage
Safe storage locker for trained models and datasets with backup protection
vLLM Inference
Production-ready AI service optimized for speed with REST API integration
Frequently Asked Questions
Will my work be lost if I logout?
No. Everything is saved automatically. Your models and data are stored safely. When you login again, everything is still there.
Can other students see my work?
No. Each student has a completely isolated workspace. Your projects and data remain private.
What if I run out of GPU memory?
The system handles this automatically. If needed, just restart your kernel and try again with smaller settings.
How long does model training take?
Usually 5-15 minutes depending on data size. Simple experiments are faster, complex ones may take longer.
Can I download my trained models?
Yes! You can download models from the model storage dashboard (MLflow) anytime.
What if something doesn't work?
Restart your Jupyter kernel first (Kernel → Restart). This fixes 90% of issues. Contact admin if the problem persists.
Ready to Start Your AI Journey
Your AI Centre of Excellence is fully operational and ready to use. Students can immediately begin training custom AI models, running fast inference, tracking experiments, and storing models safely. Everything is pre-configured and tested—no complex setup needed.
Login
Access your workspace at niet.thqindia.com
Create
Train and fine-tune AI models with GPU power
Deploy
Use your models in real applications
Track
Keep a track of your experiments niet.thqindia.com/mlflow
Support Resources: Check the "Troubleshooting" notebook in your workspace, restart your Jupyter kernel for most issues, or contact your instructor for assistance. Administrators can review the admin dashboard and Neysa Velocis Dashboard for system status.
Happy experimenting! Your journey into artificial intelligence starts now.
©2026 Technology-HQ.com. All rights reserved.