Some resources to get started with data science & AI/ML (attn: IWC DCO aspirants)

[link6]

BLUF: A short list of extremely low-barrier resources to help applicants who would like to strengthen their IWC DCO eligibility by developing data-science and AI/ML familiarity and skills.



tl;dr: Hi all,

As you're likely aware, for better or for worse data science and AI/ML are regarded as "desired" skills for a few IWC designators. I had zero knowledge of these domains until I began a graduate program in a notably "techie" lab 4 years ago, so I appreciate how arcane and inaccessible they can seem at a glance. I personally wasn't sure where to go to get hands-on experience, what material would be too daunting at the outset, or whether these fields are even amenable to self study.

Here, I've compiled a handful of resources that made these domains fun and tractable when I was initially learning the ropes, and which I hope will allow you to dip your toe in the pool before jumping in (e.g., enrolling in a pricey course, or embarking on a comp-sci degree). These are free or "freemium," presuppose no (or minimal) knowledge of computer programming, and require no installs.

Courses:

Google's Machine Learning Crash Course (https://developers.google.com/machine-learning/crash-course): Teaches key concepts at the heart of ML, like gradient descent, to comp-sci newcomers.

Codecademy Python 3 (https://www.codecademy.com/learn/learn-python-3): A gentle, hands-on intro to the Python programming language. An on-ramp to coding and not a data-science or AI/ML course per se. Freemium content, but I highly recommend just sticking with the free content.

DataCamp's Data Science Track (https://learn.datacamp.com/career-tracks/data-science-for-everyone): Loads of short (e.g., 2- to 4-hour), gamified courses packaged together as a (very big) meta-course. I highly recommend the early modules, which are excellent introductions to data science in general, as well as Python's NumPy, Matplotlib, and Pandas libraries specifically—essential stuff for data scientists. Freemium content here (at $25/mo subscription) is probably worth it if you plan to tackle several courses in short order, but there's a decent amount of free content if you're on a budget.

By working your way through ^^these, you'll quickly become conversant in key AI/ML concepts, like reinforcement learning and gradient descent, and you'll gradually develop usable Python skills centered on the most essential (and extremely transferable) scientific libraries.

Daily Practice:

Edabit Python (https://edabit.com/challenges/python3): A community site with thousands of short Python exercises, sortable by difficulty. Each problem has solutions, commentary, etc. (Like Codecademy, not a data science and AI/ML-focused experience, but a place for succinct, quality practice in syntax and algorithmic thinking, which are key enablers.)

Resources:

Python for Beginners: (https://wiki.python.org/moin/BeginnersGuide/NonProgrammers): A repository of open-source books, courses, etc. to help anyone get started.

Happy learning!

About me: I am an IWC DCO applicant for the March 2021 cycle (designators 1835 and 1815). As a student, I use Pythonic data-science and AI/ML methods in my research, but I'm absolutely still learning this stuff and am a relative newcomer to these domains. I'd love to hear your favorite low-barrier resources for getting started.

 

[fieldrat]

The AI/ML stuff is of particular interest to the 1815 crowd; immediate relevancy. Being conversant is a good breakout.

The "flex" is articulating not just why, but specific scenarios of how AI/ML can be applied to Navy IWC issues/problems. Think strategic, operational, and tactical, in that order.

 

[johngonavy]

BLUF: A short list of extremely low-barrier resources to help applicants who would like to strengthen their IWC DCO eligibility by developing data-science and AI/ML familiarity and skills.

View attachment 29712

tl;dr: Hi all,

As you're likely aware, for better or for worse data science and AI/ML are regarded as "desired" skills for a few IWC designators. I had zero knowledge of these domains until I began a graduate program in a notably "techie" lab 4 years ago, so I appreciate how arcane and inaccessible they can seem at a glance. I personally wasn't sure where to go to get hands-on experience, what material would be too daunting at the outset, or whether these fields are even amenable to self study.

Here, I've compiled a handful of resources that made these domains fun and tractable when I was initially learning the ropes, and which I hope will allow you to dip your toe in the pool before jumping in (e.g., enrolling in a pricey course, or embarking on a comp-sci degree). These are free or "freemium," presuppose no (or minimal) knowledge of computer programming, and require no installs.

Courses:



By working your way through ^^these, you'll quickly become conversant in key AI/ML concepts, like reinforcement learning and gradient descent, and you'll gradually develop usable Python skills centered on the most essential (and extremely transferable) scientific libraries.

Daily Practice:

Resources:


Happy learning!

About me: I am an IWC DCO applicant for the March 2021 cycle (designators 1835 and 1815). As a student, I use Pythonic data-science and AI/ML methods in my research, but I'm absolutely still learning this stuff and am a relative newcomer to these domains. I'd love to hear your favorite low-barrier resources for getting started.

Great but does it provide any credential the board will recognize?

 

[nodropinufaka]

The AI/ML stuff is of particular interest to the 1815 crowd; immediate relevancy. Being conversant is a good breakout.

The "flex" is articulating not just why, but specific scenarios of how AI/ML can be applied to Navy IWC issues/problems. Think strategic, operational, and tactical, in that order.

But when is a Navy officer going to be given opportunity to use these?

Sure we can talk all day about AI and ML and potential problems and how to solve with it. But when is someone who is an Officer going to be able to implement it? The contract and JAIC is with BAH and they are doing it. Not junior or mid level officers.

To the OP and everyone else. Get good at the basics. AI and ML are not the basics. Be good at telling decision makers what they need to know.

The old INDOPACOM J3 said it best "I am going to make a decision with 10 percent of the information or 100 percent of the information. Help me make the best decision"

 

[Hair Warrior]

There is a ton more AI/ML than just at the JAIC.

That said, it’s a very small niche and not the core skill for a naval intelligence officer (which is: maritime opintel).

 

[nodropinufaka]

There is a ton more AI/ML than just at the JAIC.

That said, it’s a very small niche and not the core skill for a naval intelligence officer (which is: maritime opintel).

I am not debating that there isn't more.

I am saying that you will likely not be doing it as an officer.

 

[link6]

To the OP and everyone else. Get good at the basics. AI and ML are not the basics. Be good at telling decision makers what they need to know.

(Been away for a while so apologies for the late reply...)

AI/ML is arguably an emerging basic skill. In any case—whether you aspire to be basic, intermediate, or advanced—it's an explicitly desirable skill. Since the threshold of useful competency is honestly pretty low, I thought I'd share some free(/"freemium") resources here that could help people get started if they'd like to lean forward.

 

[link6]

Great but does it provide any credential the board will recognize?

I'm not sure, tbh. I don't have a degree in CS, nor any alphabet soup-type certs. My experience is probably 75% OJT, 15% goofing around on my own, and 10% coursework under ambiguously-named classes like "Graduate Programming Seminar." I do, however, use and develop AI/ML methods on a daily basis, so my resume and personal statement leaned heavily on that. My interviewers overwhelmingly centered our conversations on my ideas about how AI/ML might benefit the IWC—something that I thought carefully about in advance of our discussions. Each interviewer expressed high AI/ML-related praise in their write-ups. At least one interviewer was a bona fide AI/ML SME; being able to speak a common dialect with him was a boon to my general poise and confidence, no doubt.

All that said, for all I know the board could have disregarded the interviewers' feedback, searched my transcripts and employment history for stuff with "AI/ML" in the title, and—finding nothing—weighted my AI/ML experience at zero. ¯\(ツ)/¯

 

[link6]

The AI/ML stuff is of particular interest to the 1815 crowd; immediate relevancy. Being conversant is a good breakout.

The "flex" is articulating not just why, but specific scenarios of how AI/ML can be applied to Navy IWC issues/problems. Think strategic, operational, and tactical, in that order.

100% agree. Even if you never intend to write a line of code, it seems increasingly urgent to understand where and (roughly) how AI/ML solutions can be implemented at various apertures.

 

[taxi1]

This book is free and yet awesome. Math foundations for statistical learning.

An Introduction to Statistical Learning

www.statlearning.com

Also great freebie, higher math.

Elements of Statistical Learning: data mining, inference, and prediction. 2nd Edition.

 

[Flash]

The AI/ML stuff is of particular interest to the 1815 crowd; immediate relevancy. Being conversant is a good breakout.

The "flex" is articulating not just why, but specific scenarios of how AI/ML can be applied to Navy IWC issues/problems. Think strategic, operational, and tactical, in that order.

But when is a Navy officer going to be given opportunity to use these?

Sure we can talk all day about AI and ML and potential problems and how to solve with it. But when is someone who is an Officer going to be able to implement it?

As @fieldrat says, being aware of what AI/ML is and how it relates to the Navy and military will be a good skill to have going forward for some officers. You aren't going to be doing the grunt work as an officer but just like not actually turning the wrenches on a platform or handling a weapon having a good knowledge of the tools you are employ is a foundational requirement as a naval officer.

 

[IKE]

Chiming in for two reasons:

1. Agree a little with both sides above. No, the (U)RL officer is unlikely to actually do any of the technical work, but there's a reason Test Pilot School (usually) only takes STEM majors with good grades, and it has nothing to do with us actually engineering anything. Being conversant, especially to the point of being to call BS in a technical realm is indispensable IMHO.

2. I'm in my first semester of a distance MS in Operations Research at Ga Tech. If anyone finds this thread and wants to know how they treat active duty using GI bill and/or program details, feel free to PM.

 

[Hair Warrior]

Interesting. I was just asked yesterday if the Navy had ORSAs. I said it wasn’t an officer designator and was probably an NOBC or AQD. I looked it up and I was right. I assume you are LS9 for now, and will be 280 AQD when you graduate. Do you have a utilization tour in mind once you’re done with your degree?

 

[IKE]

Interesting. I was just asked yesterday if the Navy had ORSAs. I said it wasn’t an officer designator and was probably an NOBC or AQD. I looked it up and I was right. I assume you are LS9 for now, and will be 280 AQD when you graduate. Do you have a utilization tour in mind once you’re done with your degree?

I haven't looked into getting an AQD, but I know there's a subspec code for the degree. I'm about to transition from one restrictive career path (1310) to another (1510). Job prospects for my next tour include: a program office, a program office, or ... wait for it... a program office. While I'm sure there are some AEDO billets managing or work beside analysts, it's not the active-duty folks doing the work.

If I could find my way into a fellowship or whatever it's called with CNA, I'd jump on it, but AEDOs have detailers and a golden path too. If I had to guess, most of the people with the AQD or doing Ops analyst jobs failed to screen for CO (like me) but are still URL (unlike me), and probably got their MS from NPS post-DH.

I'm pursuing the degree because I'm fairly certain it's the work I want to do after my AD time ends. It's also a requirement if I want to pursue a PhD (and a good way of figuring out if I really want to do that to myself and my family).

 

[Hair Warrior]

Aye. Your SSP is 321X. Once your MS is complete, you'll gain (or have to submit for) the 280 AQD, i.e. nugget ORSA. Once your utilization tour is complete, you'll gain (or have to submit for) the 281 AQD.

Ref: https://www.mynavyhr.navy.mil/Porta... Manual I 73.pdf?ver=rpCPz8Vt4b52x7ETh2GP8A==

Do you know if you are a 3210/3211 "regular" ORSA, 3212 logistics, or 3213 energy?