Fascinating story! Thank you for setting candid and authentic tone, Miguel! It is quite reflective of why I felt something genuine, something special yet real when I ran into your and Paul’s projects couple of weeks ago.
My story is long but I will try to condense it. My applied math education coincided with the sunset of last AI winter when I was enamored with the promise of AI expert systems and ran into Geoff Hinton’s early musings about its reboot. That was late 80s. AI under-delivery torpedoed all kinds of desires even to think about Hinton’s sanity :-)
Fast-forwarding some 30+ years to the end of 2022 definitely felt like vindication to Hinton’s and many others’ vision. I was so happy to go back to where I started getting glimpses of the promise so long ago.
Since last AI winter to its current renaissance I dedicated my professional life to things revolving around distributed systems. While being passionate about cutting edge technology I credit to my early mentors to think along practical value of what I do: Why->What->How->Impact.
I changed many jobs, ranging from early hot startups to established enterprises. Not because I was hopeless opportunist but because I was honest with myself - what worked and not for me, calibrating my sweet spot, constantly reinventing myself.
Two years ago I realized lot of promise lies ahead thanks to Deep Learning and decided to help people gain solid intuition about it. That’s how I started my site https://www.enterai.world and regularly published newsletters and educational posts on LinkedIn.
Too much hype, too many things are sensationalized and portrayed as voodoo. I applaud your initiative to carve out quiet and productive corner to learn building what matters - our future. Thank you for that! Will be honored to be a part of it. Eagerly looking for soulmates.
Special appreciation for quality, depth and breadth of PhiloAgent like projects. Hats off to you and Paul!
Man, your story is amazing. So you were one of the few supporters of Hinton's work at the beginning, right? That's crazy. Really happy you enjoyed PhiloAgents! And I have a question for you. Since you were working on this field during the previous AI Winter ... what's your take of our present / future? Are we heading to another one?
Dude, those were crazy “inquisition” like times 🤣 I was on verge of finishing my thesis and had to sprinkle in Expert Systems accolades. But I hated it (because rule base systems would not scale and I hated it) and hopelessly fell in love with elegance of distributing differential credits in nascent backpropagation then sheepishly advocated by a non-name Hinton (needless to say I was dreaming differentials. I know, weird me :-))
DL and LLM machinery is scientifically very sound and super potent. I am extremely bullish how far (at least from utility perspective) it will take us. Of course, there is/will be a lot of hype and distraction. But it is ok. Worth tolerating (=ignoring) it.
Take MCP. It is a much needed (and inevitable) “headless”-LLM interface standardization. Yet some “influencers” are portraying it as a God-sent Nirvana. Although it is some conceptual variation of RESTful API. But no! It is “AI Native, and hence undisputed winner”. Whatever… if we can ignore this hyperbole and focus on building useful stuff regardless acronyms - we are golden!
"Yet some “influencers” are portraying it as a God-sent Nirvana"
Is pure gold xD.
That's the reason why I'm preparing another 2h video / course about MCP + A2A. With no hype. Just build something with these two protocols in a Docker Compose app I guess. Let's see how it goes!
That kind of grounding is much needed. Make no mistake, MCP-A2A cocktail is much needed to illustrate the *inevitable* concept. Now what will be shelf-life of it? Will it evolve into something different? Who knows. But it is worth investing in conceptualizing in tangible way. You and Paul have proven to do it exceptionally well.
Keep being DL Enlightener! We are here to help moving needle in the right direction ;-)
This is so nice, both you and Paul inspire me deeply. I'm a civil engineering graduate currently navigating my way into the world of AI. It's encouraging to see others who've successfully transitioned into this field. Thank you for sharing your journey!
Wow, little bit as my history. I was in my first year of physics when, after talking to a lecturer, he presented me with job opportunities after graduation. Nothing appealed to me. I changed my major to cybernetics in economics, which involved processing huge amounts of statistical data and making predictions based on that data. My professional history carried me a bit in a different direction, but for the past 6 years I have renewed my interest in this field. First ML and now neural networks. Now I regularly meet with a group of enthusiasts and developers in this field.
It was like the next step in econometrics. When you use data to predict market behavior. Not in the classical way, computing huge data matrices, but that was the early stage of ML, like linear regression or knn. But there was no Pythorch then, so it was really a huge amount of work.
Sounds amazing. And I guess you applied notions from Physics? Like Statistical Mechanics, etc? I never got quite into econophysics, etc. But I heard Physicists were very valuable profiles in these kind of jobs. I don't know if that's true! xD
Only mathematical analysis and algebra came in handy. However, at that time, processing such an amount of data was terribly tedious and gave uninteresting results. There was a shortage of good data. Therefore, life steered in another direction. Now this field is much more interesting,and the old experience is very useful. Especially for input data processing.
Great story it is really inspirational. First, I would like to congratulate you on your channel. It is great and insightful.
I´m still trying to transition from my current position to a job in tech where I could create, learn and grow. My story is a bit chaotic to be honest and I´m still trying to pinpoint in what I should focus on to make myself a valuable asset, so I´m open to any suggestions.
I graduated from law school and specialized in the Ecuadorian Taxation Regime. Nevertheless, I pursued an accounting and finance degree in a local college. There I took linear algebra wherein the teacher gave us extra points if we used math lab to resolve homework. It was in that moment when I knew that formal logic and programming is what I really enjoy. It also made me remember back to elementary school where I excelled at Scratch 😊, but it never occur to me I could pursue something related to tech.
In that college I took Java Programming, including programming 101, OOP and Data structures. But couldn´t finish my degree.
Afterwards I developed some blockchain solutions for some friends using node js, react and the solidity programming language, by following online tutorials. Unfortunately, the project never took off so I had to maintain my position as a tax attorney.
Later I learned how to develop basic Ai solutions by using Hugging Face, Langgraph and function definitions to improve technical translations in my job. Likewise, I used such technologies to try to automate the revision of technical reports that must be filed before our tax authority. I also used PostgreSQL to try to automate trademark renewal process (and I´m trying to integrate to a AI agent to further automate the workflow).
I also learned how to use Pytorch and even installed Llama locally in my pc. I used it with CUDA since I have a Nvidia gaming gpu :D
So currently I´m trying to finish an online master’s degree in Data Science and attempting to practice, in my free time, everything related to full stack solutions (which is a lot). Since I learned python using data camp as well as in my master´s degree, I´m trying to learn Django (for front end), PostgreSQL for backend, attempting to dive deep in LLM and AI base solutions. Nevertheless, it is hard to focus on something specific that I know I could leverage. Then I found your channel and lessons where I found interesting full stack solutions I can learn from.
Fascinating story! Thank you for setting candid and authentic tone, Miguel! It is quite reflective of why I felt something genuine, something special yet real when I ran into your and Paul’s projects couple of weeks ago.
My story is long but I will try to condense it. My applied math education coincided with the sunset of last AI winter when I was enamored with the promise of AI expert systems and ran into Geoff Hinton’s early musings about its reboot. That was late 80s. AI under-delivery torpedoed all kinds of desires even to think about Hinton’s sanity :-)
Fast-forwarding some 30+ years to the end of 2022 definitely felt like vindication to Hinton’s and many others’ vision. I was so happy to go back to where I started getting glimpses of the promise so long ago.
Since last AI winter to its current renaissance I dedicated my professional life to things revolving around distributed systems. While being passionate about cutting edge technology I credit to my early mentors to think along practical value of what I do: Why->What->How->Impact.
I changed many jobs, ranging from early hot startups to established enterprises. Not because I was hopeless opportunist but because I was honest with myself - what worked and not for me, calibrating my sweet spot, constantly reinventing myself.
Two years ago I realized lot of promise lies ahead thanks to Deep Learning and decided to help people gain solid intuition about it. That’s how I started my site https://www.enterai.world and regularly published newsletters and educational posts on LinkedIn.
Too much hype, too many things are sensationalized and portrayed as voodoo. I applaud your initiative to carve out quiet and productive corner to learn building what matters - our future. Thank you for that! Will be honored to be a part of it. Eagerly looking for soulmates.
Special appreciation for quality, depth and breadth of PhiloAgent like projects. Hats off to you and Paul!
More about me at https://www.enterai.world/about and http://www.linkedin.com/in/nicoskek/
Man, your story is amazing. So you were one of the few supporters of Hinton's work at the beginning, right? That's crazy. Really happy you enjoyed PhiloAgents! And I have a question for you. Since you were working on this field during the previous AI Winter ... what's your take of our present / future? Are we heading to another one?
Dude, those were crazy “inquisition” like times 🤣 I was on verge of finishing my thesis and had to sprinkle in Expert Systems accolades. But I hated it (because rule base systems would not scale and I hated it) and hopelessly fell in love with elegance of distributing differential credits in nascent backpropagation then sheepishly advocated by a non-name Hinton (needless to say I was dreaming differentials. I know, weird me :-))
DL and LLM machinery is scientifically very sound and super potent. I am extremely bullish how far (at least from utility perspective) it will take us. Of course, there is/will be a lot of hype and distraction. But it is ok. Worth tolerating (=ignoring) it.
Take MCP. It is a much needed (and inevitable) “headless”-LLM interface standardization. Yet some “influencers” are portraying it as a God-sent Nirvana. Although it is some conceptual variation of RESTful API. But no! It is “AI Native, and hence undisputed winner”. Whatever… if we can ignore this hyperbole and focus on building useful stuff regardless acronyms - we are golden!
Man, I agree 100% with everything you said. This:
"Yet some “influencers” are portraying it as a God-sent Nirvana"
Is pure gold xD.
That's the reason why I'm preparing another 2h video / course about MCP + A2A. With no hype. Just build something with these two protocols in a Docker Compose app I guess. Let's see how it goes!
That kind of grounding is much needed. Make no mistake, MCP-A2A cocktail is much needed to illustrate the *inevitable* concept. Now what will be shelf-life of it? Will it evolve into something different? Who knows. But it is worth investing in conceptualizing in tangible way. You and Paul have proven to do it exceptionally well.
Keep being DL Enlightener! We are here to help moving needle in the right direction ;-)
Never too late.. I'm finally here. Installed substack and you were the first person I started following. Thanks for sharing your story 👍
Inspirational story, man 🤟You inspired me to do mine as well 😂 Love this personal format.
thanks! Oh yes, looking forward to that article ;)
Hehe, man your story was quite a run! I should do mine :))
P.S These long-form authentic articles suit you well, you should do more like these - keeps readers hooked
thanks man! Yes, I feel comfortable with this style :) (and ofc, we all want to hear Alex R's story!)
This is so nice, both you and Paul inspire me deeply. I'm a civil engineering graduate currently navigating my way into the world of AI. It's encouraging to see others who've successfully transitioned into this field. Thank you for sharing your journey!
thanks!! Really happy you find my journey inspirational. You need to tell us yours!!
Wow, little bit as my history. I was in my first year of physics when, after talking to a lecturer, he presented me with job opportunities after graduation. Nothing appealed to me. I changed my major to cybernetics in economics, which involved processing huge amounts of statistical data and making predictions based on that data. My professional history carried me a bit in a different direction, but for the past 6 years I have renewed my interest in this field. First ML and now neural networks. Now I regularly meet with a group of enthusiasts and developers in this field.
Yes! Very similar to mine! "Cybernetic in economics", can you elaborate on that? I'm really curious
It was like the next step in econometrics. When you use data to predict market behavior. Not in the classical way, computing huge data matrices, but that was the early stage of ML, like linear regression or knn. But there was no Pythorch then, so it was really a huge amount of work.
Sounds amazing. And I guess you applied notions from Physics? Like Statistical Mechanics, etc? I never got quite into econophysics, etc. But I heard Physicists were very valuable profiles in these kind of jobs. I don't know if that's true! xD
Only mathematical analysis and algebra came in handy. However, at that time, processing such an amount of data was terribly tedious and gave uninteresting results. There was a shortage of good data. Therefore, life steered in another direction. Now this field is much more interesting,and the old experience is very useful. Especially for input data processing.
I see. I need to investigate it. Sounds really cool, to be honest.
I love this story Miguel. I am currently at the very early stage of building but reading your blog inspired me to keep building.
Great story it is really inspirational. First, I would like to congratulate you on your channel. It is great and insightful.
I´m still trying to transition from my current position to a job in tech where I could create, learn and grow. My story is a bit chaotic to be honest and I´m still trying to pinpoint in what I should focus on to make myself a valuable asset, so I´m open to any suggestions.
I graduated from law school and specialized in the Ecuadorian Taxation Regime. Nevertheless, I pursued an accounting and finance degree in a local college. There I took linear algebra wherein the teacher gave us extra points if we used math lab to resolve homework. It was in that moment when I knew that formal logic and programming is what I really enjoy. It also made me remember back to elementary school where I excelled at Scratch 😊, but it never occur to me I could pursue something related to tech.
In that college I took Java Programming, including programming 101, OOP and Data structures. But couldn´t finish my degree.
Afterwards I developed some blockchain solutions for some friends using node js, react and the solidity programming language, by following online tutorials. Unfortunately, the project never took off so I had to maintain my position as a tax attorney.
Later I learned how to develop basic Ai solutions by using Hugging Face, Langgraph and function definitions to improve technical translations in my job. Likewise, I used such technologies to try to automate the revision of technical reports that must be filed before our tax authority. I also used PostgreSQL to try to automate trademark renewal process (and I´m trying to integrate to a AI agent to further automate the workflow).
I also learned how to use Pytorch and even installed Llama locally in my pc. I used it with CUDA since I have a Nvidia gaming gpu :D
So currently I´m trying to finish an online master’s degree in Data Science and attempting to practice, in my free time, everything related to full stack solutions (which is a lot). Since I learned python using data camp as well as in my master´s degree, I´m trying to learn Django (for front end), PostgreSQL for backend, attempting to dive deep in LLM and AI base solutions. Nevertheless, it is hard to focus on something specific that I know I could leverage. Then I found your channel and lessons where I found interesting full stack solutions I can learn from.
Many thanks for your post!
Keep up with the great content it really helps 😊