1404-06-21

Date: 1404-06-21

Time: 10:00 PM to 11:00 PM

Facilitator: Mehrdad

📌 Agenda

Greetings

Greetings

Future Directions (ML & Order Legacy)

Progress Review

ML group progress so far:
- Set up the GitHub repo
- Completed Zarchi statistics course (with notes & QAs)
- Completed two chapters of Hands-On (slower pace, but necessary for setup)
- Reflection: overall positive progress
Next Steps & Issues
- Issue 1: Different aspirations and outcomes
  - Academic focus
    - Supports future lessons
    - Informs choice of major in upcoming academic periods
    - Opens research and future job opportunities
  - Career/Job focus
  - Project-oriented focus (startup)
Discussion
- Acknowledge possibility of parting ways
  - Consider how far we can progress together
  - Explore continuing as different groups, each advancing in their own way
  - Maintain communication to support mutual progress
- Review aspirations:
  - Identify common ground
  - Identify key differences
ML Jobs: Q&A with Mohsen
- How much time is needed before being job-ready?
- What should be done to prepare?
- Should we continue reading Hands-On and advance ML knowledge?
- Should we focus on projects and engineering practice?
- If aiming for resume/GitHub projects:
  - What do you recommend?
  - What mini-projects can we do for both learning and resume building?
Resources & Books
- How different is job readiness compared to academic readiness?
- What resources?

Next week tasks

Specifying reading and practical tasks for next week

📝 Notes & Discussion

Notes on ML Job Readiness & Career Paths Based on Mohsen’s Experience

Most jobs today involve some form of machine learning, often with a bit of deep learning sprinkled in. However, the level of required expertise varies a lot depending on the role.

Applied ML/AI Roles
Many companies mainly want people who can use AI rather than deeply develop it. These positions require basic ML knowledge but rely more on engineering skills: integrating APIs, deploying pre-trained models, handling data pipelines, and making systems reliable.
Domain-Specific Roles (e.g., Computer Vision)
Other companies, particularly in areas like computer vision, expect stronger technical depth. Here, it’s important to understand data preparation, be familiar with different model architectures, and have the ability to read and apply research papers. These positions lean heavily toward R&D, so they demand a solid academic and technical background.
Core ML/Deep Learning Roles
Jobs that focus purely on machine learning and deep learning — for example, handwriting OCR or building custom neural architectures — require a very high level of knowledge. These roles expect mastery of theory, math, and implementation details.
MLOps / AIOps Roles
Another branch of careers lies in ML operations: ensuring models run reliably at scale, managing infrastructure, monitoring performance, and automating deployment. This requires less theory-heavy ML knowledge and more focus on engineering and systems.

👉 Takeaway: For “AI application” jobs, a broad but shallow understanding might be enough. For research- or development-focused jobs, significantly deeper knowledge is needed.

Resume & GitHub Project Recommendations

If you want to build projects to showcase on your resume and GitHub, there are a few different directions:

AI Application / RAG Systems
- Example: a Retrieval-Augmented Generation (RAG) system, potentially backed by something like an MCP server.
- These are more advanced but very impressive for a portfolio. They demonstrate system integration, prompt engineering, and applied ML knowledge.
Computer Vision Projects
- Lots of well-defined tasks exist here: image classification, object detection, segmentation, etc.
Core ML / Neural Network Projects
- Much harder to do without strong background knowledge.
- Options include implementing well-known architectures, replicating papers, or joining Kaggle competitions.
- Kaggle is useful, but the competitions are often highly competitive.

Academic ML Recommendations

Focus Areas

Emphasis on mathematical foundations: linear algebra, probability, statistics, optimization.
Stronger theoretical depth than job-ready tracks, preparing for research, advanced courses, and R&D roles.

Suggested Starting Points

An Introduction to Statistical Learning (ISLR)

Accessible entry point, with practical applications (mainly in R, but concepts generalize).
Free official PDF: ISLR Book
Pattern Recognition and Machine Learning (Christopher Bishop, 2006)
More advanced than ISLR, with heavier probability and Bayesian methods.
Great for academic study and research grounding.
The Elements of Statistical Learning (Hastie, Tibshirani, Friedman, 2009)
Companion to ISLR but more mathematically rigorous.
Useful once ISLR concepts are clear.
Free official PDF: ESL Book

Other Recommended Resources

Scikit-Learn Documentation

User Guide – clear tutorials on classical ML algorithms.
Complements theoretical study with practical Python examples.
Mathematics for Machine Learning (Deisenroth et al., 2020)
Free Book covering linear algebra, probability, and optimization specifically for ML learners.
Deep Learning (Goodfellow, Bengio, Courville, 2016)
Book link – standard academic reference on deep learning theory.

Final Notes

No matter which path we take — academic, job-oriented, or project-based — we all need a solid foundation in ML concepts. With that shared base, each of us can progress toward our own goals, while still supporting each other along the way. Building this foundation will take a few months of steady effort.

How We Continue

Each person should share their aspirations (academic, job, or project) along with the pros and cons they see.
Over time, it’s natural that our paths may diverge:
- Some may lean toward academic research.
- Some may pursue projects/startups.
- Others may focus on jobs or applied ML.
That’s okay. We can collaborate now, learn together, and later form smaller groups aligned with similar goals — while keeping communication open between all groups.

Applied ML & Documentation

If some decide to start applied ML projects early, we should:

Document the process carefully.
Create a dedicated space for each project.
Advance them in parallel while maintaining shared discussions.
This way, projects become both learning opportunities and resources for others.

Our Conclusion (Best Path for Now)

Learn ML Academically
- Provides depth for future lessons.
- Helps in deciding academic majors.
- Builds a strong theoretical foundation for research or advanced roles.
Build Resume & GitHub Projects
- Develop a few mini-projects (treat them like homework).
- Showcase them on GitHub for credibility and future job applications.
Flexible Course Approach
- Use Hands-On Machine Learning as a guiding reference.
- Allow flexibility: people may choose different online courses/resources.
- Each person can watch/read their preferred material, then share notes and insights with the group.

✅ Decisions

Shared Foundation
- Everyone will build a strong foundation in ML concepts (math, theory, and practical understanding) over the next few months.
Paths & Aspirations
- Each member will clarify their primary aspiration:
  - Academic focus
  - Job/engineering focus
  - Project/startup focus
- People with similar goals may form subgroups to collaborate more closely.
Course & Learning Approach
- Mandatory: Each session will include Hands-On Machine Learning study and at some mini-project / practice.
- Mini-projects can be added to GitHub and used for resumes or learning purposes.
- Optional: Members can choose additional courses/resources individually (e.g., ISLR, Scikit-Learn tutorials, Kaggle competitions) based on personal interest or focus.
- Group discussions will focus on sharing learnings, insights, and practical tips from both the mandatory Hands-On sessions and any optional resources chosen by members.
Applied ML & Mini-Projects
- Members who want to do applied ML projects now can:
  - Document their process
  - Share progress and insights with the group
Future Collaboration
- Maintain communication across subgroups, even if paths diverge.
- Periodically check in to see what everyone is learning, sharing resources, and potentially collaborating on larger projects in the future.

Tasks

Task Description	Assigned To	Related Topic	Due Date
Read [Hands-On ML] 3rd edition: Chapter 3 (https://github.com/ageron/handson-ml3)	Everyone	Classification	1404-06-28
AIC Quera Questions 1. Pandas 2. LinearAlgebra-Statistics-Probability 3. Linear Regression 4. Polynomial Regression	Everyone	Math Foundation Python Basics Intro to ML	1404-06-28

🔁 Next Session

Proposed date: TBD
Next topic: Classification
Facilitator: TBD

📎 Resources Mentioned

Textbooks
- Pattern Recognition and Machine Learning (Bishop, 2006)
  Classic academic reference, heavy on probability and math foundations.
- An Introduction to Statistical Learning with Applications in R (ISLR, 2nd Edition, 2021)
  Accessible, practical introduction with R examples (free PDF available).
- The Elements of Statistical Learning (Hastie, Tibshirani, Friedman, 2nd Edition, 2009)
  More advanced than ISLR, deeper theory. Free PDF available.
Documentation
- Scikit-Learn Documentation
  Official docs with tutorials, API reference, and practical examples.
Math Foundations
- Linear Algebra (MIT OpenCourseWare, Gilbert Strang)
- Mathematics for Machine Learning (free book by Deisenroth et al.)
Machine Learning Courses

Course	Description	Link
Zarchi ML Course (Ali Sharifi-Zarchi, Sharif University of Technology)	An undergraduate course offering free access to slides, Jupyter notebooks, videos, exercises. Offered by the Computer Engineering department. (GitHub)	Sharif ML Course GitHub (GitHub)
IBM Machine Learning Course / Professional Certificate	A multi-course specialization on Coursera by IBM. Covers supervised & unsupervised learning, regression, classification, clustering, deep learning, reinforcement learning, hands-on labs & projects. (Coursera)	IBM Machine Learning Professional Certificate on Coursera (Coursera)
Andrew Ng Machine Learning Course / Specialization	Beginner-friendly program (by DeepLearning.AI + Stanford Online) covering foundational ML concepts, visual intuitions, code examples, etc. Updated “Machine Learning Specialization”. (DeepLearning.AI)	Machine Learning Specialization by Andrew Ng (DeepLearning.AI)

Machine Learning

Explorer