ISO 9001:2015 & ISO 27001:2022 Certified

AI Technoverse Infotech Pvt. Ltd.

Training Programme

AI Course in Vashi

Two levels, one path. Start with Python, LLM APIs and your first RAG application, then move to agents, fine-tuning and production AI on AWS and Azure.

  • Two Levels
  • 32 & 48 Hrs
  • Live Projects
  • Certificate

Download the Brochure

Ask about this programme

Which level suits you, what the batches cost and when the next one starts — answered by the trainer, not a call centre.

We use these details only to answer your enquiry.

01 — Overview

Artificial Intelligence Training Institute in Vashi with Live Projects

Our Artificial Intelligence Course in Vashi, Navi Mumbai runs in two levels and is taught by engineers who build AI systems for clients. The Basic course takes you from Python foundations through machine learning concepts, LLM APIs, prompt engineering and your first working RAG application. The Professional course then moves you into production territory: advanced retrieval, multi-agent systems, fine-tuning open-source models, and deploying on AWS and Microsoft Azure.

Every module carries a hands-on lab, and both levels finish with a live project you build, deploy and present. You leave with a public GitHub repository and a running application rather than a folder of notes. Take either level on its own, or work through both in sequence.

02 — Two Levels

Start from Python, or straight into production AI.

Take either level on its own, or work through both in sequence.

BASIC · Basic

AI Foundations

8 Weeks · 32 Hours · 2 live projects

Designed for beginners with no prior AI experience. Covers Python essentials, core AI and machine learning concepts, working with LLM APIs, prompt engineering, and building and deploying a real RAG application.

  • Set up a Python AI environment and work with the OpenAI, Anthropic and Gemini APIs
  • Apply professional prompt engineering to get reliable, structured output
  • Build a working RAG pipeline with LangChain, ChromaDB and an LLM API
  • Deploy an AI web application using FastAPI and Streamlit
  • Understand embeddings, vector search and document chunking strategies

PRO · Professional

Advanced AI

12 Weeks · 48 Hours · 3 live projects + 1 capstone

Builds on the Basic course. Covers advanced RAG architectures, production vector databases, AI agents, open-source model fine-tuning, AWS AI services, Microsoft Azure AI, and full production deployment with MLOps practices.

  • Design production-grade RAG systems with HyDE, re-ranking and hybrid search
  • Build multi-agent AI systems using LangGraph, CrewAI and AutoGen
  • Fine-tune open-source LLMs such as LLaMA 3 and Mistral using LoRA and QLoRA
  • Deploy AI on AWS — Bedrock, SageMaker, Lambda, Rekognition and Textract
  • Deploy AI on Azure — Azure OpenAI, AI Studio, Azure ML and AI Search
  • Ship production systems with Docker, GitHub Actions CI/CD and LangSmith monitoring
Course Brochure

The full syllabus in one PDF.

The brochure carries the complete module list for both levels, the lab and project breakdown, the tool set and the certification details, in one file you can share with a manager or a parent.

  • Basic — 7 modules across 32 hours
  • Professional — 10 modules across 48 hours
  • Live projects, tools and certification

Download the brochure

Tell us where to send it and the brochure lands in your inbox, with a download link here too.

We use these details only to send the brochure and answer your enquiry.

03 — Course Content

What each level covers.

Basic Course — AI Foundations

8 Weeks · 32 Hours · 7 modules · 8 hands-on labs

M01 Python for AI
  • Python environment setup — VS Code, Jupyter, virtual environments
  • Core Python — data types, functions, loops and object-oriented programming
  • Working with JSON, CSV and file input/output
  • NumPy arrays and Pandas DataFrames for data handling
  • Consuming REST APIs with the requests library
  • Lab — a Python script that turns a raw dataset into AI-ready output
M02 AI & Machine Learning Fundamentals
  • AI, machine learning, deep learning and generative AI — the real differences
  • Supervised, unsupervised and reinforcement learning
  • Neural networks — layers, weights and activation functions
  • How models learn — loss functions, gradient descent and backpropagation
  • Transformer architecture and the attention mechanism, explained simply
  • Tokens, embeddings, context windows, temperature and top-p
  • Lab — visualise a forward pass and compare model outputs across parameters
M03 Large Language Models & APIs
  • OpenAI API — chat completions, system and user roles, streaming, rate limits
  • Anthropic Claude API — messages format, system prompts, long context windows
  • Google Gemini API — multimodal text and image inputs
  • Cost management — token tracking and model tiering
  • Secrets management with environment variables and python-dotenv
  • Lab — a command-line chatbot that switches between multiple LLM providers
M04 Prompt Engineering
  • Zero-shot, one-shot and few-shot prompting with worked examples
  • Chain-of-thought and tree-of-thought prompting
  • Role prompting, persona prompting and contextual framing
  • Structured outputs — JSON and schema-based responses validated with Pydantic
  • System prompt design and prompt chaining
  • Hallucinations, prompt injection and jailbreak risks — and how to guard against them
  • Lab — a prompt template library covering eight professional use cases
M05 Embeddings & Vector Databases
  • How text becomes vectors, and what semantic similarity actually measures
  • OpenAI Embeddings API, cosine similarity and dot product
  • FAISS — local vector indexing and querying
  • ChromaDB — persistent vector store, collections and metadata
  • Chunking documents, attaching metadata and top-k retrieval
  • Lab — a semantic search engine over your own document set
M06 RAG — Retrieval Augmented Generation
  • Why RAG exists — hallucinations and knowledge cutoffs
  • RAG versus fine-tuning — choosing the right approach
  • Pipeline architecture — ingest, chunk, embed, store, retrieve, generate
  • Document loaders for PDFs, text, CSV and web pages
  • Chunking strategies — fixed size, recursive and overlapping
  • LangChain — chains, retrievers, prompt templates and RetrievalQA
  • Adding conversation memory to a RAG chatbot
  • Lab — a RAG chatbot answering questions over documents you upload
M07 Deployment & Live Project
  • FastAPI — REST endpoints that call LLMs, with Pydantic schemas
  • Streamlit — chat interfaces, file upload and session state
  • Cloud deployment and an introduction to Docker containers
  • Live project — a complete RAG application for a domain you choose
  • GitHub repository with README and architecture documentation
  • Live demo and project walkthrough

Tools and technologies

  • Python 3.11+
  • Jupyter Notebook
  • VS Code
  • Git & GitHub
  • OpenAI API
  • Anthropic API
  • Google Gemini API
  • python-dotenv
  • NumPy
  • Pandas
  • FastAPI
  • LangChain
  • ChromaDB
  • FAISS
  • Streamlit
  • Docker (intro)

Professional Course — Advanced AI

12 Weeks · 48 Hours · 10 modules · 12 hands-on labs

M01 Advanced Python & System Design for AI
  • asyncio fundamentals — event loop, coroutines, async and await
  • Concurrent LLM API calls using asyncio and aiohttp
  • Pydantic v2 — validation, custom validators and nested models
  • Logging, error handling and retry logic for AI systems
  • Configuration management, packaging and unit testing with pytest
  • Lab — a production-ready async module wrapping three LLM providers
M02 Advanced Prompt Engineering & Function Calling
  • ReAct prompting — combining reasoning and acting
  • Prompt compression and context window optimisation
  • Self-consistency and ensemble prompting
  • OpenAI function calling and Anthropic tool use — defining and parsing tools
  • Giving models controlled access to external APIs and databases
  • Guardrails — input and output validation, content moderation, JSON mode
  • Lab — an LLM that calls real tools including search, calculator and weather APIs
M03 Advanced RAG Architectures
  • Semantic, parent-child and sliding-window chunking
  • HyDE — hypothetical document embeddings for better retrieval
  • Multi-query retrieval and query expansion
  • Re-ranking with cross-encoders, Cohere Rerank and ColBERT
  • Hybrid search — combining BM25 keyword and vector retrieval with RRF
  • Evaluation with RAGAS — faithfulness, answer relevancy, context precision
  • Lab — an advanced RAG pipeline with HyDE and re-ranking over 1000+ documents
M04 Production Vector Databases
  • Pinecone — namespaces, metadata filtering, upserts and serverless
  • Weaviate — hybrid search, multi-tenancy and schema design
  • Qdrant — payload indexing, quantisation and performance tuning
  • pgvector — using PostgreSQL as a vector store
  • Choosing a vector database — cost, latency and scalability trade-offs
  • Lab — migrate a RAG pipeline from ChromaDB to Pinecone with metadata filtering
M05 AI Agents & Orchestration
  • Agent architecture — reasoning loops, tool use, memory and planning
  • How agents differ from chains and pipelines
  • LangGraph — stateful multi-step workflows, conditional edges, human-in-the-loop
  • CrewAI — role-based agents and task delegation
  • Microsoft AutoGen — conversational multi-agent systems
  • Agent memory — short-term, long-term and episodic
  • Lab — a multi-agent research pipeline: researcher, analyst, writer, reviewer
M06 LlamaIndex & Open-Source LLMs
  • LlamaIndex versus LangChain — when each is the right choice
  • Data connectors and index types, query engines and recursive retrieval
  • Hugging Face Hub and running LLaMA 3, Mistral and Phi-3 locally with Ollama
  • vLLM for high-throughput inference; GGUF and GPTQ quantisation
  • Fine-tuning with LoRA and QLoRA using the PEFT library
  • When to fine-tune instead of using RAG
  • Lab — fine-tune Mistral or LLaMA 3 on a domain-specific dataset using QLoRA
M07 AWS AI Services
  • IAM, Amazon S3 and AWS Lambda foundations for AI workloads
  • Amazon Bedrock — model access, Converse API, Knowledge Bases and Agents
  • Amazon SageMaker — Studio, training jobs, endpoints, Canvas and Pipelines
  • Amazon Rekognition — image and video analysis with custom labels
  • Amazon Comprehend — entity recognition, sentiment and classification
  • Amazon Transcribe, Textract and Lex
  • Lab — a document classification pipeline across S3, Lambda, Bedrock and Comprehend
M08 Microsoft Azure AI Services
  • Resource groups, Azure Active Directory identities and Blob Storage
  • Azure OpenAI Service — deployments, private endpoints, RBAC and content filtering
  • Azure AI Studio — visual prompt flow, model catalog and evaluation
  • Azure Machine Learning — AutoML, pipelines and managed endpoints
  • Azure AI Vision, Speech, Language and Document Intelligence
  • Azure AI Search — hybrid retrieval and semantic ranking in RAG pipelines
  • Lab — a RAG chatbot on Azure OpenAI, AI Search and App Service
M09 MLOps & Production Deployment
  • FastAPI — streaming responses, async endpoints and background tasks
  • Docker and Docker Compose for multi-container AI systems
  • Container registries — Docker Hub, AWS ECR and Azure Container Registry
  • CI/CD with GitHub Actions — test, build, push and deploy
  • Observability with LangSmith and Langfuse
  • Monitoring cost, latency and error rates; caching and model tiering
  • Lab — a production AI API deployed with Docker and GitHub Actions CI/CD
M10 Capstone Project & Certification
  • A full-stack AI application solving a real business problem
  • Must combine RAG or agents with an LLM API, a vector database and cloud deployment
  • Choose your own domain — legal, healthcare, finance, HR, education or e-commerce
  • GitHub repository with clean code, README and architecture diagram
  • A deployed, publicly accessible FastAPI or Streamlit interface
  • Benchmarking report covering latency, accuracy and cost per query
  • Technical demo and question-and-answer session with an instructor panel

Tools and technologies

  • Python 3.11+
  • Pydantic v2
  • asyncio
  • pytest
  • FastAPI
  • Uvicorn
  • Docker
  • OpenAI API
  • Anthropic API
  • Google Gemini API
  • Hugging Face
  • Ollama
  • vLLM
  • LangChain
  • LangGraph
  • LlamaIndex
  • CrewAI
  • AutoGen
  • Pinecone
  • Weaviate
  • Qdrant
  • pgvector
  • ChromaDB
  • FAISS
  • Amazon Bedrock
  • Amazon SageMaker
  • Amazon Rekognition
  • Amazon Comprehend
  • AWS Lambda
  • Azure OpenAI
  • Azure AI Studio
  • Azure ML
  • Azure AI Search
  • LangSmith
  • Langfuse
  • RAGAS
  • PEFT / LoRA
04 — Live Projects

You finish with something that runs.

Every module carries a hands-on lab, and each level ends in a project you build, deploy and present.

BASIC · 2 live projects

Basic Course

8 hands-on labs

  • RAG chatbot answering questions over your own uploaded documents
  • Capstone — a complete RAG application for a chosen domain such as HR, legal, finance or education, deployed with a documented GitHub repository

PRO · 3 live projects + 1 capstone

Professional Course

12 hands-on labs

  • Multi-agent research pipeline built with LangGraph and CrewAI
  • Document classification pipeline on AWS using S3, Lambda, Bedrock and Comprehend
  • RAG chatbot deployed on Azure OpenAI with AI Search and App Service
  • Capstone — a full-stack production AI application with a benchmarking report and panel demo
05 — Who It Is For

Built for people who want to ship AI, not just discuss it.

The Basic course assumes no prior AI experience and starts from installing Python.

  • Students and fresh graduates

    Final-year and recent graduates in IT, computer science or engineering who want AI skills an employer can actually test.

  • Software developers

    Working developers adding LLM APIs, RAG and agent frameworks to what they already build, without leaving their current role.

  • Data and analytics professionals

    Analysts and data engineers moving from reporting into machine learning, embeddings and applied generative AI.

  • Business owners and managers

    Decision makers who need to judge what AI can genuinely do for their operation before committing budget to it.

06 — FAQ

Artificial Intelligence Course in Vashi

Do I need programming experience to join the AI course?

Not for the Basic course. Module 1 starts with installing Python and works up from data types and functions, so you can join without having written code before. What you do need is willingness to practise between sessions, because the labs build on each other. The Professional course is different and assumes you are already comfortable writing Python.

What is the difference between the Basic and Professional course?

The Basic course is 32 hours across 7 modules and takes you to the point where you can build and deploy a working RAG application using LLM APIs. The Professional course is 48 hours across 10 modules and moves into production work: advanced retrieval, multi-agent systems, fine-tuning open-source models, and deploying on AWS and Azure with proper MLOps practice.

Do I have to complete the Basic course before the Professional one?

Not necessarily. If you already write Python confidently and have used an LLM API, we can assess you straight into the Professional course. If either of those is shaky, start with the Basic course, because the advanced track assumes that grounding and moves quickly.

How many live projects will I work on?

The Basic course includes 2 live projects, ending with a complete RAG application for a domain you choose. The Professional course includes 3 live projects plus a full capstone: a multi-agent research pipeline, an AWS document classification pipeline, an Azure RAG chatbot, and then a production application you scope, build, deploy and present. Every module also carries its own hands-on lab.

Do I need my own laptop?

Yes. Bring a machine with at least 8 GB of RAM, 60 GB free disk space and administrator rights so you can install Python and the tooling. You do not need a GPU — the fine-tuning work in the Professional course runs on free cloud GPU environments, and cloud accounts for the AWS and Azure modules are set up during the sessions.

Will I actually learn AWS and Azure, or just hear about them?

You will build on both. The Professional course dedicates a full module to each, and both end in a deployed lab — a document classification pipeline on AWS, and a RAG chatbot on Azure. The Basic course covers both platforms at overview level so you understand what is available before you commit to one.

Who will be the trainer?

A working AI engineer from our services team rather than a full-time lecturer. The people building retrieval systems and agent workflows for clients take the sessions, which is why the frameworks and model versions taught stay current with what is actually being used.

Is the training available online?

Yes. Every session runs as a live online cohort as well as an offline batch at our Vashi centre, and a hybrid option is available. Online learners do the same labs on their own machines with the trainer on the call.

What do I get at the end?

A certificate for the level you complete, issued by AI Technoverse Infotech Pvt. Ltd., plus the project work itself. That means a public GitHub repository, a deployed application and, for Professional students, a benchmarking report — the things an interviewer can actually look at.

What are the fees and how do I enrol?

Send an enquiry using the form on this page or call us. We will share the current fee for each level, the next batch dates and the payment options, and tell you honestly which level you should start at.

Not sure which level?Our IT counsellors are here to help.

Tell us what you already know and where you want to end up. We will point you at the level that fits, even if that means starting lower than you hoped.

Get in Touch
Brochure