Live Lecture 1​

Overview

The session was an extensive Q&A during a cohort orientation for a generative AI program.
Participants asked questions about course structure, tracks (entrepreneurship vs.
employment), execution paths (code vs. no-code), hardware requirements, scheduling,
career outcomes, and more. The instructors and mentors shared advice, clarified doubts,
and offered guidance on how to get the most out of the course.

Key Topics & Insights

●​ Tracks & Execution Paths

○​ Tracks:
■​ Entrepreneurship Track: For those looking to launch products or
startups.
■​ Employment Track: For those aiming to secure a job in the AI industry.
■​ Note: Once you choose a track, switching mid-course is discouraged.
○​ Execution Paths:
■​ Code Path: Recommended if you have coding experience or want to
deepen your technical skills.
■​ No-Code Path: Ideal for non-coders or those who prefer rapid product
development without deep coding.
○​ Dedicated sessions (on March 24th and 26th) will help you decide which track
fits your background and goals.
●​ Course Components
○​ Live Lectures & Recordings:
■​ Scheduled at 6:00 PM IST for about two hours, with recordings
available later on the LMS (Learning Management System).
○​ Office Hours & Mentorship:
■​ Personalized sessions to help you with project work, track-specific
queries, and career guidance.
○​ Assignments & Capstone Projects:
■​ Projects are a core part of the course and must be completed (three
capstone projects in total) to be considered as having finished the
program.
○​ Bounties & Networking:
■​ Bounty challenges offer real-world problem statements with cash
prizes.
■​ Networking sessions and community channels (like Discord) are
available for collaboration and career opportunities.
○​ Additional Resources:
■​ Cloud-based GPU access is provided, so you don’t need high-end
hardware.
 ■​ Supplementary coding materials (e.g., CS50, Python refreshers) are
recommended if you opt for the code track.
●​ Advice & Best Practices
○​ Commitment:
■​ Regular attendance at live lectures, active participation in office hours,
and engaging with the community are critical.
○​ Learning Approach:
■​ Focus on building a solid foundation in problem solving and
computational thinking.
■​ If you're unsure which track to pick, attend the dedicated sessions and
speak with mentors (e.g., via Discord or scheduled discovery calls).
○​ Career Outcomes:
■​ The program emphasizes outcomes—whether you’re looking to build
a product or land a job.
■​ Success stories from past cohorts illustrate that both coders and
non-coders can achieve their goals, and many former students have
been hired by partnering companies.
●​ Miscellaneous Q&A Highlights
○​ Hardware & Tools:
■​ A standard computer (even an M1/M2 MacBook) is sufficient since
computation is cloud-based.
○​ Time Commitment:
■​ Expect to invest a few hours per day (an extra 30–60 minutes on top
of live sessions if you choose the code path).
○​ Job & Startup Guidance:
■​ Whether you are aiming for a career change or want to launch a
startup, the program offers tailored support (e.g., resume help,
mentorship, and connections with venture studios).

Conclusion

The session emphasized that the success of the cohort lies in the commitment and active
participation of each student. The instructors encouraged everyone to take full advantage of
live sessions, office hours, and networking opportunities while also sharing practical
tips—from choosing the right track to leveraging community resources for career growth.

Overall, the key takeaway is to trust the process, commit to learning, and use the available
resources to solve real-world problems with generative AI.
Live Lecture 2 - History of Gen AI and Tensor Art

1. Welcome & Introduction

●​ Greetings & Setup:​

The instructor welcomes everyone to Lecture 1 of the Applied AI Cohort, checks
audio/visual settings, and outlines the session.
●​ Lecture Focus:​
The session covers a brief history of AI, an introduction to diffusion models via a tool
called Tensor Art, and a deep dive into “control nets” for guiding image generation.

2. A Brief History of AI & Generative Models

●​ Early AI & Computation:

○​ Discusses AI’s origins during WWII (Alan Turing and the Enigma code) and
the development of early neural networks inspired by the human neuron.
○​ Mentions the Turing Test and early models (like the perceptron) that set the
stage for AI.
●​ Key Milestones & Hype Cycles:
○​ Highlights breakthroughs like Deep Blue defeating Kasparov, Watson winning
Jeopardy!, and DeepMind’s achievements.
○​ Explains how AI has gone through multiple “hype cycles” and “AI winters”
before reaching practical applicability today.
○​ Emphasizes that modern advances are driven by exponential improvements
in computing (Moore’s Law) and open-source foundational models.

3. Introduction to Tensor Art & Diffusion Models

●​ What Is Tensor Art?

○​ An online tool for experimenting with image generation using diffusion
models.
○​ It offers a simple UI with prompt input, model selection (e.g., Flux, Stable
Diffusion 3, SDXL), and adjustable parameters like image dimensions and
sampling steps.
●​ Hands-On Demonstration:
○​ The instructor walks through generating images (e.g., a bird on a tree branch
or a futuristic high-tech computer) by entering prompts and tweaking settings.
○​ Emphasizes that while basic settings are shown today, more advanced
parameters (like guidance scales, upscaling, and sampling methods) will be
covered in future sessions.
4. Deep Dive into Control Nets

●​ Purpose of Control Nets:

○​ Control Nets allow you to guide and refine generated images beyond a
simple text prompt.
○​ They help impose structure such as specific poses (open pose), depth
(foreground vs. background), or edge details (canny edges).
●​ Demonstration & Usage:
○​ The instructor demonstrates using control nets to overlay a desired pose or to
add artistic details to an image.
○​ Adjustments like the “strength” of the control (typically set around 0.6–0.7)
ensure the desired output while maintaining flexibility.
○​ Several types are mentioned (open pose, depth, canny) along with tips on
combining them for more control.

5. Q&A Highlights & Practical Guidance

●​ Interactive Questions:
○​ Students ask about using multiple control nets, achieving consistent
characters, refining details (e.g., facial features), and safety concerns when
using personal images.
○​ Answers stress that while the tool currently offers 512×512 resolution
(upgradable to 1024×1024), experimenting is key.
○​ Additional resources (such as the "three blue one brown" YouTube series) are
recommended for deeper theoretical understanding.
●​ Assignments & Resources:
○​ Practice Sets:
■​ Optional exercises include generating images using various control
nets and reverse-prompting an existing image.
■​ Students are encouraged to share their outputs on Discord for peer
and mentor feedback.
○​ Staying Updated:
■​ Follow curated Twitter accounts, GitHub Trending pages, and provided
links (via the LMS “Wise” platform) to keep up with the fast-evolving AI
landscape.
○​ Next Steps:
■​ The upcoming lectures will delve into the theory of diffusion and
practical sessions using tools like ComfyUI.

6. Conclusion & Final Remarks

●​ Takeaway Message:
○​ Treat Tensor Art as a “training wheel” for understanding diffusion models.
 ○​ Embrace experimentation and immerse yourself in learning—even if it feels
overwhelming at first.
●​ Wrap-Up:
○​ The session ends with a reminder to complete the practice sets, stay active
on Discord, and check LMS for additional materials and transcripts.
○​ The instructor signs off, promising that the next lectures will build on today’s
concepts with more hands-on and theoretical insights.