These are all amazing projects, looking forward to learning from them!
Two open source tools I use in my MLOPs work day to day on a personal hobby basis are DVC and CML, check them out and see where you can incorporate them.
Check out my repo here where I used them in a Gitops workflow to finetune llama2 on a self hosted Github Actions runner with just 8GB of VRAM:
Hey Miguel love the ideas of what we can learn. I have a question. Am right now wondering if I must continue my studies in data science and Ai (formal degree in a university) or dedicate myself to this type of training that is emerging a lot from highly skilled and experience professionals like you. My debate is overall because I found myself with very bad teachers even though it opened my mind to new ideas and concepts. What do you recommend me? (Money is a factor is kind of concerning that you pay a lot in university but not get the amount of knowledge and experience) by the way am 40
I think it really depends on what kind of learning you're looking for. Universities are great at building strong theoretical foundations, but when it comes to teaching the practical skills you need to actually land a job… they tend to fall pretty short 😅
If you've already started a degree, I'd recommend finishing it—it gives you a solid theoretical background that many people are missing these days. You can then complement that with cohorts like this one, which focus on applying those concepts in practice to solve real-world problems (the ones companies actually care about!).
Ohh okay, so I see you are offering cohort based courses and I see many courses available on substack. I am new to your publication and I want to get started right now but clueless from where to start. I am working as a risk analyst (data science) but would like to get into building Agents and in general want to keep myself up for the trends in AI. So, what do you recommend me how do I get started with your publication?
Congrats man! Your free courses were already close to cohorts, engineering first, and packed with insights. Excited to see what you'll bring up next ;)
Unreal article and I'm sure cohorts is something that a lot of people are going to appreciate specifically someone who is trying to upgrade their career, their skills, and move forward with their career as well!
Hi Miguel ...what does the paid subscription comprise and does it provide Responsible AI engineer types of projects and how to do on a no code platform or what language do you use. How much is paid subscription please ?? Is the work standard across all LLMs??
Your emphasis on 'real-world engineering over hype' is exactly the mindset shift that made my AI workflows actually productive.
I've been building with multiple AI models in 2026 and the interesting thing is: different models excel at different tasks. Claude for complex reasoning, GPT for creative writing, Gemini for research. The challenge is orchestration - deciding which model handles what.
The MLOps skills you're covering (CI/CD, model registries, drift monitoring) are underrated. When I built Wiz - my personal AI agent system - the infrastructure decisions mattered as much as the prompts.
I completely agree with your perspective on traditional ML systems. The ability to design and build them is often underestimated today, which is exactly why we plan to dive deeply into this area in 2026!
Looks good! Whatever topic you decide to go with, I think what matters is the production level quality. That, and systems level thinking.
Absolutely! Systems-level thinking is one of the key areas we'll be focusing on in 2026.
These are all amazing projects, looking forward to learning from them!
Two open source tools I use in my MLOPs work day to day on a personal hobby basis are DVC and CML, check them out and see where you can incorporate them.
Check out my repo here where I used them in a Gitops workflow to finetune llama2 on a self hosted Github Actions runner with just 8GB of VRAM:
https://github.com/ShahNewazKhan/skynet
I think you can incorporate DVC and CML into the traditional ML content you are making.
Thanks for sharing this! Yes, I was fully planning to incorporate DVC, and I'll also review CML.
Super excited for this awesome journey 🎉 😁
Just one question
For cohort, the paid substack will be enough or do we need to enroll separately?
Cohorts will run separately, but paid Substack members will always receive big discounts + priority access to seats!
Awesome thank you 😁
Hey Miguel love the ideas of what we can learn. I have a question. Am right now wondering if I must continue my studies in data science and Ai (formal degree in a university) or dedicate myself to this type of training that is emerging a lot from highly skilled and experience professionals like you. My debate is overall because I found myself with very bad teachers even though it opened my mind to new ideas and concepts. What do you recommend me? (Money is a factor is kind of concerning that you pay a lot in university but not get the amount of knowledge and experience) by the way am 40
I think it really depends on what kind of learning you're looking for. Universities are great at building strong theoretical foundations, but when it comes to teaching the practical skills you need to actually land a job… they tend to fall pretty short 😅
If you've already started a degree, I'd recommend finishing it—it gives you a solid theoretical background that many people are missing these days. You can then complement that with cohorts like this one, which focus on applying those concepts in practice to solve real-world problems (the ones companies actually care about!).
Thanks for always taking the time to reply!
Hi Miguel 😊
I would like to have a 1:1 with you to get my journey started. I am not sure how I can book a session with you. Can you help me with that?
Thanks,
Chirag
Hey! The 1:1 sessions haven’t been officially launched yet. Once they’re available, I’ll make an announcement!
Ohh okay, so I see you are offering cohort based courses and I see many courses available on substack. I am new to your publication and I want to get started right now but clueless from where to start. I am working as a risk analyst (data science) but would like to get into building Agents and in general want to keep myself up for the trends in AI. So, what do you recommend me how do I get started with your publication?
Congrats man! Your free courses were already close to cohorts, engineering first, and packed with insights. Excited to see what you'll bring up next ;)
Unreal article and I'm sure cohorts is something that a lot of people are going to appreciate specifically someone who is trying to upgrade their career, their skills, and move forward with their career as well!
thanks!! Glad you liked it :)
If possible, please share some insights on the ML roadmap and specialization in LLM, NLP, and computer vision.
Sure, will do in the future ;)
Also Harvard just recently open sourced their ML System Book / Course
https://github.com/harvard-edge/cs249r_book
The road map they lay in there is quiet nice!
thanks for sharing this!!
Hi Miguel ...what does the paid subscription comprise and does it provide Responsible AI engineer types of projects and how to do on a no code platform or what language do you use. How much is paid subscription please ?? Is the work standard across all LLMs??
Your emphasis on 'real-world engineering over hype' is exactly the mindset shift that made my AI workflows actually productive.
I've been building with multiple AI models in 2026 and the interesting thing is: different models excel at different tasks. Claude for complex reasoning, GPT for creative writing, Gemini for research. The challenge is orchestration - deciding which model handles what.
The MLOps skills you're covering (CI/CD, model registries, drift monitoring) are underrated. When I built Wiz - my personal AI agent system - the infrastructure decisions mattered as much as the prompts.
For anyone interested in how different AI models work together in practice, I wrote about my multi-model workflow: https://thoughts.jock.pl/p/multi-model-ai-workflow-2026-gpt-claude-gemini
I completely agree with your perspective on traditional ML systems. The ability to design and build them is often underestimated today, which is exactly why we plan to dive deeply into this area in 2026!