VECTOR Institute

Generative AI &
Agentic Systems

Machine Learning to LLMs, RAG and AI agents —
built, deployed and production-ready!

Eligibility

Students & Graduates

Duration

4 Months

Program Overview

From Machine Learning and Deep Learning through NLP, Transformers and Large Language Models to Generative AI, RAG and Agentic AI systems, deployed with MLOps and finished with an industry-grade capstone project.

Generative AI & Agentic Systems program overview

Who This Program Is For

  • 2nd, 3rd and Final-Year Engineering Students
  • Fresh Graduates
  • Anyone passionate about building a career in Artificial Intelligence

Prerequisites

  • Python programming and basic data analysis with NumPy and Pandas
  • Logical thinking and willingness to learn are essential

Learning Outcomes

  • Hands-on projects in Machine Learning, Deep Learning, NLP, Generative AI and Agentic AI — from a YOLO object detector to a production RAG system, multi-agent systems and a custom MCP server
  • Develop an industry-grade end-to-end capstone project, from problem identification to deployment

Online & Classroom Training

Choose how you learn: in the classroom at our Hyderabad centre, or live online from anywhere in India. Both modes follow the same 4-month syllabus, with the same faculty and the same hands-on projects.

Classroom Training — Hyderabad

  • Faculty-led classes at our Hyderabad centre
  • Theory and hands-on practical sessions in every module
  • Lab practice on every topic, with the faculty at hand to clear your doubts
  • Project guidance and mentoring from the faculty
  • Learn alongside other students in a focused classroom environment

Live Online Training

  • Live instructor-led classes in secured Zoom sessions, with the same faculty as the classroom
  • Classes at the same time as the Hyderabad batch: ask questions and clear doubts in real time
  • Dedicated Learning Management System (LMS) for course content and assignments
  • The same syllabus, projects and assessments as the classroom
  • Placement assistance: online students take the Placement Eligibility Test at the institute

Upcoming Batches

LocationModeDateStart Time
New batch dates will be announced soon. Please call us or send an enquiry.

Program Modules

Module 1. Machine Learning — Core Algorithms

Scikit-learn · Regression · Trees · SVM · Clustering · Evaluation

Supervised Learning

  • Linear & Logistic Regression
  • Ridge & Lasso (Regularization)
  • K-Nearest Neighbors (KNN)
  • Decision Trees & Random Forests
  • Gradient Boosting (XGBoost)
  • LightGBM
  • CatBoost
  • Support Vector Machines (SVM)
  • Naive Bayes classifier
  • Ensemble methods

Unsupervised Learning

  • K-Means Clustering
  • Hierarchical Clustering
  • DBSCAN (density-based)
  • PCA — Dimensionality Reduction
  • t-SNE visualization
  • Anomaly Detection methods

Model Engineering

  • Bias-Variance tradeoff
  • Cross-validation strategies
  • Feature selection & importance
  • Hyperparameter tuning (GridSearch)
  • Evaluation metrics
  • Model interpretability (SHAP, LIME)
  • Imbalanced datasets (SMOTE)
  • Pipeline building
  • Saving & loading models
  • Hands-on ML Mini Projects

What You Will Build

  • End-to-end pipeline: data prep → feature engineering → XGBoost → evaluation report
  • Customer segmentation system using K-Means and PCA with interactive Plotly visualizations
  • Anomaly detection engine for financial transactions using DBSCAN and Isolation Forest
  • Scikit-learn ML pipeline with GridSearch tuning, cross-validation and model serialization

Outcome: Confidence to build, evaluate, and explain any ML model — in interviews and in production

Module 2. Deep Learning & Computer Vision

TensorFlow / PyTorch · CNNs · Transfer Learning · Object Detection

Neural Network Foundations

  • Neurons & network architecture design
  • Activation & loss functions
  • Optimizers — Adam, SGD, RMSProp
  • Backpropagation & gradient flow
  • Overfitting, dropout & regularization
  • Batch normalization

Computer Vision

  • CNNs — Conv, Pool, Flatten layers
  • Transfer learning (ResNet, EfficientNet)
  • Object detection with YOLO

What You Will Build

  • Custom CNN trained from scratch on a multi-class image classification dataset
  • Transfer learning pipeline using ResNet / EfficientNet for domain-specific image recognition
  • Real-time object detection system powered by YOLO on live video or image input

Outcome: Build real-world image-based AI systems — using state-of-the-art architectures

Module 3. NLP, Transformers & Large Language Models

GPT · HuggingFace · Sentiment · Prompt Engineering

NLP Fundamentals

  • Text preprocessing & tokenization
  • Bag-of-Words & TF-IDF
  • Word2Vec embeddings
  • Sentiment analysis
  • Named Entity Recognition (NER)
  • Text classification

Sequence Models

  • RNN, LSTM, GRU architectures
  • Bidirectional RNNs
  • Encoder-Decoder architecture
  • Attention mechanism deep dive

Transformers & LLMs

  • Transformer architecture — self-attention, multi-head attention, positional encoding, encoder, decoder, residual connections, layer normalization, feed-forward networks, masking, tokenization, context windows
  • BERT
  • GPT series — architecture & usage
  • HuggingFace pipelines & models
  • Prompt Engineering strategies
  • Embeddings & vector search
  • Gemini / Claude / GPT API integration
  • Chatbot development with LLMs

What You Will Build

  • Sentiment Analysis using TF-IDF, Word Embeddings, and LSTM
  • AI Chatbot using GPT / Gemini / Claude APIs with Prompt Engineering
  • Semantic Search Application using Embeddings and Vector Search

Outcome: Master the fundamentals of Transformers and Large Language Models to build production-ready NLP applications, semantic search systems, and AI-powered chatbots

Module 4. Generative AI, RAG & Agentic Systems

LangChain · LangGraph · Agents · Vector DBs · MCP · A2A

LLM Application Foundations

  • LangChain fundamentals
  • Prompt Templates
  • Chains & LCEL (LangChain Expression Language)
  • Output Parsers & Structured Outputs
  • Streaming
  • LangSmith (tracing & debugging)
  • Model selection strategies
  • Token counting & cost optimization

Context Engineering

  • Data ingestion & preprocessing
  • Document chunking strategies
  • Metadata enrichment
  • Embeddings & semantic search
  • Vector databases (FAISS, Chroma, Pinecone)
  • Context window management
  • LLM application architecture

Retrieval-Augmented Generation (RAG)

  • RAG system design & architecture
  • Retrieval techniques & hybrid search strategies
  • Query understanding & query rewriting
  • Prompt engineering for RAG
  • Memory & conversational RAG
  • Self-RAG & Graph RAG
  • Agentic RAG
  • RAG evaluation metrics

Agentic Systems — Core Concepts

  • Foundations of Agentic AI
  • Core Agent Architecture
  • Tool Use / Function Calling — external APIs, Python tools, database tools, web search tools, calculator tools, file system tools
  • Planning & Reasoning
  • Memory Systems
  • Multi-Agent Systems

Orchestration Frameworks

  • LangGraph — graph architecture, state management, nodes, edges, conditional routing, cycles, checkpointing
  • LangChain Agents
  • CrewAI

Agent Communication Protocols & Standards

  • MCP (Model Context Protocol)
  • A2A (Agent-to-Agent protocol)
  • Function / tool schema standards
  • Agent discovery mechanisms

Autonomous Agent Design Patterns

  • Supervisor pattern
  • Planner–Executor pattern
  • Hierarchical agents
  • ReAct
  • Tree of Thoughts
  • Reflection & self-consistency
  • Planning loops
  • Deep Research agents
  • Human-in-the-Loop
  • Agent streaming

Evaluation, Safety & Observability

  • Prompt evaluation
  • Agent evaluation
  • RAG evaluation
  • Tracing, logging & monitoring
  • AI Guardrails & Safety
  • Responsible AI

What You Will Build

  • Production RAG system: PDF ingestion → ChromaDB / FAISS → semantic retrieval → grounded LLM response
  • Autonomous research agent using LangChain + tool calling (web search, calculator, database lookup)
  • Multi-agent orchestration system using LangGraph with stateful memory and task delegation between agents
  • CrewAI orchestration system: role-based automated content pipeline with agent personas (Researcher, Writer, Editor)
  • Human-in-the-Loop (HITL) workflow: high-stakes execution agent built with LangGraph
  • Deep Agent planner: complex reasoning agent with Tree of Thoughts / ReAct for multi-step research
  • Model Context Protocol (MCP) integration: custom MCP server bridging local environment with an LLM

Outcome: Design and build autonomous AI systems powering modern real-world AI products

Module 5. MLOps & Model Deployment

FastAPI · React JS · Docker · Git · Cloud · CI/CD · Monitoring

Version Control & Tracking

  • Git & GitHub
  • Branching & pull requests
  • Model registry & versioning

Model Serving & APIs

  • FastAPI endpoint design
  • REST API design principles
  • API documentation with Swagger
  • Authentication & security basics (JWT)

Frontend for AI Apps

  • JavaScript fundamentals
  • Linux Fundamentals
  • React basics for AI-powered UIs

Containerisation & CI/CD

  • Docker fundamentals
  • Dockerizing ML applications
  • GitHub Actions (CI/CD pipelines)

Cloud Deployment

  • AWS (EC2, S3, RDS)
  • Cloud deployment strategies

Monitoring & Observability

  • Application & model logging
  • Model performance monitoring & drift detection
  • Basic observability tooling (Prometheus / Grafana concepts)

What You Will Build

  • FastAPI-served ML model with JWT authentication, versioned endpoints, and Swagger documentation
  • Dockerised ML application with Docker Compose for multi-service orchestration
  • GitHub Actions CI/CD pipeline: automated testing → build → deploy on every commit

Outcome: Deploy and maintain production-grade AI models like a professional ML Engineer

Module 6. Capstone Project — Industry Grade

Data → Model → API → Dashboard → Cloud → Portfolio

Project Execution

  • Real-world problem statement selection
  • Data collection & pipeline design
  • EDA & feature engineering
  • Model building & evaluation
  • Generative AI or Agentic component
  • FastAPI backend development
  • Docker containerisation
  • Deployment to cloud platform

Career Preparation

  • GitHub portfolio setup
  • Resume preparation guidance
  • Technical interview preparation
  • Mock interviews with feedback

Capstone Deliverables

  • A GitHub repository with clean code, README, architecture diagram, and demo instructions
  • A deployed web application or API accessible via a public URL

Outcome: A job-winning, fully deployed AI portfolio project ready for top industry roles

Tools & Technologies

PythonVS CodeGoogle ColabNumPypandasMatplotlibSeabornMySQLMongoDBFAISSChromaDBscikit-learnTensorFlowPyTorchYOLONLTKspaCyHugging FaceKerasTavilyLangChainLangGraphRAGGeminiOllamaCrewAIMCPA2A

Student Testimonials

From students of Vector India’s Full Stack Development program

Have Any Questions?

1. What is the Generative AI & Agentic Systems course?

A 4-month course, in the classroom at Hyderabad or online, in six modules: machine learning, deep learning and computer vision, NLP and Large Language Models, Generative AI with RAG and AI agents, MLOps, and an industry-grade capstone project.

2. What do I need to know before joining?

Python programming and basic data analysis with NumPy and Pandas. Logical thinking and willingness to learn are essential.

3. Can I take this course online or in the classroom?

Both. You can attend in the classroom at our Hyderabad centre, or join live online classes from anywhere in India. Both modes follow the same syllabus, with the same faculty and the same projects; online classes run live at the same time as the Hyderabad batch.

4. How are the online classes conducted?

Online classes are live and instructor-led, in secured Zoom sessions with the same faculty who teach in the classroom, so you can ask questions and clear doubts in real time. A dedicated Learning Management System (LMS) gives you the course content and assignments.

5. What will I build during the course?

Hands-on projects in every module — for example a customer segmentation system, a YOLO object detector, an LLM chatbot, a production RAG system, multi-agent systems with LangGraph and CrewAI and a custom MCP server — and one industry-grade capstone project with a GitHub repository and a deployed web application or API.

6. Which Machine Learning and Deep Learning topics are covered?

Supervised learning (regression, decision trees, random forests, XGBoost, LightGBM, CatBoost, SVM), unsupervised learning (K-Means, DBSCAN, PCA, anomaly detection), model engineering and interpretability (SHAP, LIME), neural networks with TensorFlow / PyTorch, CNNs, transfer learning (ResNet, EfficientNet) and object detection with YOLO, then NLP, sequence models and Transformers.

7. Which Generative AI and Agentic AI topics are covered?

Large Language Models (GPT, Gemini, Claude) and prompt engineering, embeddings and vector databases (FAISS, ChromaDB, Pinecone), Retrieval-Augmented Generation (RAG) including conversational, Graph and Agentic RAG, AI agents with tool calling, planning and memory, multi-agent systems with LangGraph and CrewAI, the Model Context Protocol (MCP) and Agent-to-Agent (A2A) protocol, and evaluation, guardrails and responsible AI.

8. Will I learn to deploy AI models?

Yes. The MLOps module covers Git and GitHub, serving models with FastAPI (with Swagger documentation and JWT authentication), React basics for AI front ends, Docker, CI/CD with GitHub Actions, deployment on AWS (EC2, S3, RDS), and model monitoring and drift detection.

9. What is the capstone project?

An industry-grade, end-to-end project on a real-world problem: data collection and pipeline design, EDA and feature engineering, model building and evaluation, a Generative AI or agentic component, a FastAPI backend, Docker containerisation and deployment to the cloud. You finish with a GitHub repository (clean code, README, architecture diagram and demo instructions) and a deployed web application or API accessible via a public URL.

10. Do you provide placement assistance?

Yes. Placement assistance is provided to students who meet the placement eligibility criteria: at least 75% attendance and a 40% internal score in theory, lab and internal assessments, recommendation in the mock interview and communication skills assessment, and successful completion of the prescribed projects. Online students who want placement assistance must also score at least 40% in the Placement Eligibility Test, held in person at the institute.

11. How does the course prepare me for jobs?

The capstone phase includes career preparation: setting up your GitHub portfolio, resume preparation guidance, technical interview preparation and mock interviews with feedback. Your deployed capstone project and the projects you build during the course form your portfolio.

12. How do I enrol?

Admission is direct: there is no entrance test. Click Apply Now or Enquire Now on this page and fill in the form, or call us, and our team will help you with the next batch and enrolment.

13. Will I get a certificate?

Yes. At the end of the course, students receive a Merit / Course Completion Certificate, provided they meet the eligibility in terms of attendance and internal assessments.

Why Join Vector India?

Our vision is to empower the next generation of AI professionals with industry-focused education that bridges the gap between academic learning and real-world applications, developing AI engineers who build intelligent, scalable and production-ready solutions.

Apply Online Quick Enquiry Chat with us