100% found this document useful (1 vote)
126 views9 pages

Modern AI Pro Essentials

Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
100% found this document useful (1 vote)
126 views9 pages

Modern AI Pro Essentials

Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 9

MITRAROBOT.

COM
Modern AI Pro
Essentials
SIGNATURE PROGRAM FROM:
Balaji Viswanathan / Mitra Robot

Executive Summary
In the rapidly evolving landscape of technology, staying ahead requires not only
understanding the fundamentals, but also mastering practical applications of cutting-edge
innovations. The Essentials level of our Modern AI Pro bootcamp is meticulously designed to
empower mid-level managers in the tech industry with the knowledge and skills necessary to
leverage AI effectively within their organizations.

1. Objectives
1. Understand Modern AI Foundations: Grasp the current state, history, and core principles
of AI and machine learning.
2. Build POC Applications: Develop practical skills to create AI models for production using
open-source tools.
3. Data Handling and Ethics: Master data encoding, visualization, cleaning, and ethical AI
practices.
4. Experiment Management and Deployment: Manage AI experiments with MLflow and
deploy applications using Gradio.
5. Practical AI Application Skills: Set up and customize local LLMs, handle multimodal
data, and build neural networks.
6. Unstructured Data Extraction and NLP: Extract and process unstructured data and use
NLP techniques with Hugging Face.
7. Advanced AI Integrations: Build advanced chatbots, secure LLM applications, integrate
SQL databases with LLMs, and develop agent-based AI systems.

2. Target Audience
Mid-Level 1. Professionals in mid-management roles aiming to enhance AI
Managers in Tech understanding and application.
2. Managers needing to oversee and deploy AI-driven projects
within the enterprise.

Experience Level: 1. Basic Python coding experience, seeking to gain expertise in


AI and practical skills in AI applications.
2. Experience in managing technical teams would be a plus,
with a desire to acquire hands-on proficiency in AI tools.

3. Trainer Profile
Introduction: Dr. Balaji Viswanathan, CEO of Mitra Robot.
VC funded startup based in San Francisco, CA.
40+ Fortune 500 customers.
CEO Featured in Forbes, Fortune, CNN and more.

Education: MS in CS (thesis in AI Agent Systems) from Univ of Maryland.


PhD in CS with (disseration in AI/Robotics) from Univ of MD
MBA (President’s Scholar) from Babson College

Experience: 4 years as a Developer at Microsoft HQ in Redmond


2 years as Product Manager at Black Duck, Boston
8 years building AI & robotics at Mitra Robot


Won multiple
national and
international
awards for
innovations.
4. Certifications
For every session, we will conduct a quick live assessment at the end to reinforce learning
and ensure comprehension of key concepts. Additionally, there will be a 20-minute final
assessment that participants can take 1-7 days following the completion of the bootcamp.
Successful completion of these assessments will earn participants the Modern AI Pro
Essentials Certification, validating their expertise and readiness to apply AI in their
professional roles.
5. Program Benefits
Drive Innovation and Efficiency: Equip your team with cutting-edge AI tools and
techniques to enhance operational efficiency and foster innovation.
Enhance Project Outcomes: Apply practical AI skills to real enterprise challenges,
resulting in tangible improvements in project performance.
Foster Continuous Learning: Cultivate a culture of ongoing learning and adaptability
within your organization, ensuring your team stays ahead of technological
advancements.
Hands-On, Collaborative Experience: Participate in interactive coding sessions, group
activities, and real-world case studies to reinforce learning and encourage collaboration.

6. Day-wise plan
Day 0 (1.5 hours): 1. Lab session for Installation and setup of the tools.
2. Walking through the Pre-requisites

8 hours of recorded sessions with ML fundamentals. We will give a


week’s time to get this done.

Day 1 (5 hours): 1. Introduction to Modern AI and Machine Learning Fundamentals


2. Data Encoding, Visualization, Cleaning, AI Ethics, and Managing
AI Experiments

Day 2 (7.5 HRS): 1. Setting Up Local LLMs with Ollama and Prompt Engineering
2. Deep Neural Networks and Text Classification
3. Multimodal Data Handling and Image Classification
4. Unstructured Data Extraction and NLP Foundations

Day 3 (7.5 HRS): 1. Building Advanced Chatbots with Enterprise Data


2. Security of LLM Applications and Access Control
3. Combining SQL Databases with LLMs
4. Agent Architecture and Building Agents with Crew AI

7. Detailed Syllabus
Session 1: Introduction to Modern AI and ML Fundamentals
Description: This session provides an overview of the current state of AI, tracing its evolution
from early machine learning algorithms to today's advanced AI models. It includes the theory
of machine learning, concepts like train-test split, types of errors (Type 1 and 2), and
measuring accuracy.

We will also introduce you to the Generative AI Impact framework on where the key business
value in Generative AI is being generated.

Session 2: Cleaning, Visualization, Ethics, and Experiments


Description: This session covers the training models, handling real world data, cleaning and
organizing them and running your experiments.

Participants will engage in hands-on problems, such as home price prediction for regression
and income class prediction for classification. Additionally, this session covers the
importance of data encoding with label encoder vs. one-hot encoding and explores various
visualization techniques to understand data distributions and relationships. It also includes
data cleanup and imputation techniques to handle missing data, along with a discussion on
AI ethics and bias, highlighting best practices to ensure fair and unbiased models.

Participants will also learn to deploy AI applications with Gradio for interactive user
interfaces. Additionally, the session introduces Kaggle, guiding participants through setting
up an account, participating in key competitions, and leveraging the platform for learning and
collaboration.

Session 3: Setting Up Local LLMs and Prompt Engineering


Description: This session focuses on setting up local Large Language Models (LLMs) using
Ollama. We will download two state-of-the-art models (the best of that week) and run a
simple application on top of them. This will provide an introduction to building LLM
applications in an enterprise context.

We will also dive deep into prompt engineering, exploring the 12 key elements including
tone, context, instruction, and constraint. Through these activities, participants will gain a
comprehensive understanding of how LLMs work and how they can be harnessed effectively
in an enterprise setting.

Session 4: Deep Neural Networks and Text Classification


Description: Dive into the world of deep neural networks, exploring both the theoretical
foundations and practical applications. Participants will build a text classifier using Artificial
Neural Networks (ANN), gaining hands-on experience with neural network architecture and
training processes. Additionally, the session will cover essential concepts such as Word2vec
and embeddings, which are crucial for text representation and natural language processing
tasks.

Session 5: Multimodal LLMs and Image Classification


Description: This session covers the techniques for handling multimodal data, starting with
image classification using Convolutional Neural Networks (CNN). Participants will learn to
preprocess and augment image data, design and train CNNs, and evaluate model
performance. The session will also explore building multimodal models that integrate
different types of data (e.g., text and images), providing a comprehensive understanding of
how to leverage diverse data sources for more robust AI applications.

We will also bring in a couple of Multimodal LLMs such as Llava and Anthropic and talk about
their usecases.

Session 6: Unstructured Data Extraction and NLP


Description: Learn advanced techniques for extracting and processing unstructured data
from various sources, including Excel files, PDFs, webpages, audio, and video. This session
will introduce the Hugging Face ecosystem, showcasing how to use state-of-the-art natural
language processing (NLP) models for tasks such as text classification, sentiment analysis,
question answering, and summarization. Participants will gain practical skills in handling and
transforming unstructured data into valuable insights. We will also use tools such as
Unstructured.io to build complex applications that involve extracting data from tables and
images.

Session 7: Building Advanced Chatbots with RAG.


Description: Focus on the development of advanced chatbots that leverage enterprise data.
Participants will learn to extract data from PDFs, crawl websites for relevant information,
build and query vector databases, and create memory systems to enhance chatbot
performance. The session will also cover the implementation of Retrieval-Augmented
Generation (RAG) workflows, enabling the creation of chatbots that provide accurate and
contextually relevant responses.

This is the heart of this bootcamp, where we talk about key aspects of Chatbot design.

Session 8: Security of LLM Applications and Access Control


Description: This session addresses the critical aspects of securing LLM applications and
implementing effective access control measures. Participants will learn about potential
security vulnerabilities in LLM deployments and best practices to mitigate these risks.
Additionally, the session will cover the use of Langfuse for evaluating and monitoring LLM
performance, ensuring that deployed models maintain high standards of reliability and
security.

Session 9: Combining SQL Databases with LLMs


Description: Most of the structured enterprise data is stuck inside SQL databases. Explore
the integration of SQL databases with LLMs to achieve hybrid data extraction and analytics
capabilities. Participants will learn to vectorize SQL databases, enabling the use of LLMs for
complex query generation and data retrieval. This session will demonstrate how to leverage
the strengths of both SQL databases and LLMs, providing a powerful toolset for data-driven
decision-making and advanced analytics.

Session 10: Agent Architecture / Building Agents with Crew AI


Description: Understand the principles and benefits of agent-based architecture in AI
applications. This session will cover the theoretical foundations of agent-based systems and
their practical implementations. Participants will learn to build AI agents using Autogen and
Crew AI, focusing on scenarios where an agent-based approach is advantageous. The
session will highlight the flexibility and scalability of agent-based systems in handling
complex AI tasks and workflows. We will also be bringing ways to integrate real time data
into LLMs.

8. Projects
We will guide you with 5 projects based on the content provided in the course. We will
support the projects with lab sessions.

1. Project 1: Binary Classification on Titanic Disaster Data


2. Project 2: Quora Insincere Questions Classification
3. Project 3: Classify Images from CIFAR-10 Dataset
4. Project 4: Sentiment Analysis on Amazon Reviews
5. Project 5: Identifying Fake Jobs

9. Logistics
Hardware Requirements:

A good development laptop with a minimum of 8 GB RAM and 50 GB disk space.

Software Requirements:

1. Install Miniconda: Miniconda Installation Guide


Restart the terminal and create your environment: conda create -n modernai
python=3.11
2. Install Git: Git Installation Guide
Restart the terminal and run: git clone <https://github.com/balajivis/modernaipro.git>
3. Install VSCode: VSCode Installation Guide
4. Install Ollama: Ollama Download
5. Familiarize yourself with accessing the terminal/command line on your computer. For
Windows users, it will be either PowerShell or CMD.

If you have Python installed already, ensure the version is not 3.12. Use a virtual environment
or Miniconda to get a lower Python version if necessary. Contact us if you need technical
help.

User Accounts to Create:

1. Kaggle
2. Groq
3. Anthropic
4. Hugging Face
5. Langfuse

10. Conclusion
The Modern AI Pro Essentials bootcamp is more than just a training program; it's a gateway
to the future of technology. By equipping your team with cutting-edge AI skills and practical
knowledge, you are not only enhancing their capabilities but also driving your organization
towards innovation and excellence.

Our comprehensive curriculum, designed and led by Dr. Balaji Viswanathan, ensures that
participants gain a deep understanding of AI principles and hands-on experience with real-
world applications. From setting up local LLMs and prompt engineering to building advanced
chatbots and securing AI applications, this bootcamp covers all the essential aspects to
transform your team's approach to AI.

Certification and Assessments

After completing the bootcamp, participants will take a 20-minute final assessment within 1-7
days. This final assessment (80%) combined with the class assessments (20%) will
determine your final score. Participants must achieve a score of 60% or more to earn the
Modern AI Pro Essentials Certification, recognizing their expertise and readiness to apply AI
in their professional roles.

Final Thoughts
Embark on this transformative journey with us and unlock the full potential of AI for your
organization. With the guidance of industry experts and a robust support system, your team
will be well-equipped to lead AI-driven projects, innovate solutions, and make strategic
decisions that drive success. Join us in shaping the future, one AI application at a time.

You might also like