Sandboxes
7
Nuvepro’s Projects
1
Lab Type : Nuvepro’s Sandboxes
Tools Delivered : OracleDB, Chrome, Mozilla
Customer Segment : A Tier 1 Global IT Services and Consulting Organization.
Lab Type : Nuvepro’s Sandboxes
Tools Delivered : Azure Fabric Account Labs
Customer Segment : A Workforce Upskilling Accelerator providing corporate training and professional development solutions.
Lab Type : Nuvepro’s Projects
Problem Statement : This guided GenAI project is designed to provide hands-on exposure to core Generative AI concepts, workflows, and implementation approaches. The project enables learners to explore practical AI use cases, understand modern AI development workflows, and build foundational skills in applying GenAI technologies in real-world enterprise scenarios.
- Lab 1: Understanding Tokenization and Embeddings
Learn how text is split into tokens and converted into embeddings (vectors). Compare sentence similarity using embedding distance/similarity. - Lab 2: LLM Settings and Parameters
Experiment with temperature, top-p, max tokens, and penalties to understand how they affect creativity, length, determinism, and cost. - Lab 3: Exploring the OpenAI API
Make a basic API call to an LLM, inspect the JSON response, and extract the generated output for display or saving. - Lab 4: Building a Vanilla Chatbot
Build a simple multi-turn chatbot using role-based message history (system/user/assistant) to maintain conversation context. - Lab 5: Prompt Chaining with LangChain
Use LangChain chaining concepts (LCEL) to connect multiple prompts into a workflow for multi-step reasoning or basic RAG tasks. - Lab 6: Structured Outputs with LangChain
Generate validated structured outputs using schemas (Pydantic/JSON), ensuring predictable formatting and error handling. - Lab 7: Calculating Basic Evaluation Metrics
Compute precision, recall, and F1 score on a sample classification dataset and interpret what the scores mean practically. - Lab 8: Evaluating Using Custom Rubric Criteria
Evaluate LLM responses using qualitative rubrics like relevance, clarity, completeness, and tone to compare responses effectively. - Lab 9: Fine-Tuning for Classification
Prepare a labeled dataset and run a fine-tuning job for text classification, then test the tuned model on new samples. - Lab 10: Efficient Parameter-Efficient Fine-Tuning (PEFT)
Understand PEFT by tuning fewer parameters for faster, cheaper training and compare outputs against the base model. - Lab 11: RAG with LlamaIndex
Build a document index and implement retrieval-augmented generation to answer queries using only relevant document chunks. - Lab 12: Multi-Query RAG with LangChain
Improve retrieval recall by generating multiple alternate queries, comparing single-query vs multi-query retrieved contexts. - Lab 13: Building an Agentic Router
Create an intent-based router that selects the correct tool/workflow depending on the user query type. - Lab 14: Building a ReAct Agent
Implement a ReAct-style agent that reasons, takes tool actions, observes results, and generates a final grounded response.
Customer Segment : A Leading IT Training and Workforce Development Organization.
Lab Type : Nuvepro’s Sandboxes
Tools Delivered : DevOps Tools with Kubernetes
Customer Segment : A Workforce Upskilling and Emerging Technology Learning Platform.
Lab Type : Nuvepro’s Sandboxes
Tools Delivered : Azure Account Labs
Customer Segment : A Global IT Training, Certification, and Cloud Learning Solutions Provider.
Lab Type : Nuvepro’s Sandboxes
Tools Delivered : Ubuntu Server with React and ExpressJS
Customer Segment : A Leading Higher EdTech Platform providing online upskilling programs.
Lab Type : Nuvepro’s Sandboxes
Tools Delivered : Ubuntu Server with React and ExpressJS
Customer Segment : A Leading Higher EdTech Platform providing online upskilling programs.
Lab Type : Nuvepro’s Sandboxes
Tools Delivered : Azure Cloud Native Labs
Customer Segment : A Leading IT Training and Workforce Development Organization.