The inspiration and much of the material for the below GenAI resources comes from none other than Patrick Debois, the Prime Mover and OG of DevOps cultural movement and DevOpsDays.
The first NYC GenAI DevOps Hackathon was held on January 24th 2024.
Like Ghent in 2009, it was a transitional moment in DevOps: dev, sec, and ops have a cool cousin: data science and LLMOps.
Through the last 15 years there have been transformational waves in devops like DevSecOps (security invited to the party) and around 2013 global acceptance then 2016 the global hegemony of devops practices.
The addition of LLMs to integrations and new toolings feels very right. The dev/sec/ops frame Patrick provides makes the material even more accessible and understandable. As a learner, teacher/presenter, he has a special way of doing this. The magic ingredient seems a joy in learning which is also accessible and very relatable for me. Check out his learning llms repo for dev/sec/ops.
In these spaces, the true love of learning and sharing is as present as the air we breathe. It was a catalyst for a cultural movement and subsequent technological reformations. For visionaries like Dubois and Willis, the constant hunger for knowledge and the drive to share creates a massive wave of curiosity and learning. Ideas take root in the DevOps communities nurtured by an infectious enthusiasm. This continual growth cultivates a cultural movement in-situ that blossoms into groundbreaking technological advancements. One of the joyful gifts of DevOpsDays is this local context.
-
MongoDB and Amazon AWS sponsored the event. Stacks in the Hackathon included using AWS Sagemaker, Kendra, Amazon Q and S3 to create a sample app.
-
MongoDB Atlas and vector databases were a focused and using MongoDB was a highlight as easy-to-deploy. See [MongoDB setup sheet] (all free-tier).
-
Takeaway: Working with LLMs and tool integrations with the sensibilities of DevSecOps is the default. Solutions being built by startups have DevSecOps (and now LLMOps) sensibilities baked-in. The emergence of models as utility OS and the prodigious growth of and how to create value from unharnessed unstructured data.
- MongoDB as a sponser has a ton of awesome resources out there. Apparently they're on Tour and it's a cool approach for even wider exposure to MongoDB Atlas, the power of vectorstores and true engagement through DevOps OGs Patrick and John. (See separate section to highlight the MongoDB.local Sessions) This apporoach makes it super easy to get Atlas rockin, learn about vector databases and producing tangible things in hours. Pretty dope.
- Repos for the January 24th Hackathon
- Follow along and complete the steps for the Hackathon.
- Enjoy the readout from the day's session
- Base repo for OperationalizingAI
- Be sure to checkout the August 2023 GenAI workshop.
- April 8th resources and links
- April 20 (San Francisco)
- May 2nd (MongoDB.local Hackathon at the Javitz Center)
- AI and Prompt Engineering Playlists
- Operationalizing Generative AI: From ChatGPT to Production
- Video: Dev, Sec and Ops Meet Langchain
- [GenAI Tutorials for Dev, Sec, and Ops]GitHub repos of Lessons for Dev, Sec and Ops
- Using Langchain - Primers - How-Tos and Use Cases
- LangSmith
- LangServe Deploy LangChain runnables and chains as a REST API. Bananas
- Related: Willis: The LOR Stack
- Getting started building
- Examples of how to create a private LLM 1on AWS
- Reference Guide
- PromptFoo
- Fabric Prompting with templates. Daniel Miessler has interesting content worthy of my follow.
- AIConfig from LastMileAI AIConfig - the open-source framework for building production-grade AI applications. I am in love.
- PromptLayer
-
In researching the domain(s), here are some suggested reading that kept coming up that I'd like to dig into.
-
Fei-Fei Li: The Worlds I See Audible Well written and a great read/listen.
- John Willis: (Title TBD on History of Artificial Intelligence).
- Bostrom, N. (2014). Superintelligence: Paths, dangers, strategies. Oxford University Press.
- Tegmark, M. (2017). Life 3.0: Being human in the age of artificial intelligence. Knopf Doubleday Publishing Group.
- Bhaskar, S. ([2021?]). The Coming Wave.
- Chamarrow, G. I Human.
- Russell, S. J., & Norvig, P. Artificial Intelligence: A Modern Approach Pearson Education.
- Mark Hinkle, [Reuven Cohen], Patrick Debois, Joseph Enoch, John Willis, Bill Bensing, HuggingFace, Ray Myers: Youtube: Craft vs Cruft
- Timnit Gebru was terminated from Google after the publication of this paper. This was ~2021 before LLMs were as common knowledge.
-
Provide Python Primer. Covering basics like datatypes, structures, loops, conditionals. [Links] Feature using Jupyter notebooks, local and cloud based. How to launch and use Jupyter Notebooks
-
Git and version control primers and quick how-tos
-
Using APIs, step through request/response e.g. get, post, demo using Postman
-
Theoretical fundamental AI learnings and subsets. ML(subset: deep learning supervised/unsupervised), neural networks, NLP, reinforcement learning, prompt engineering, computer vision. Neural networks:review (forward and back propagation, gradient descent algorithm, weighting )
-
The difference between neural network architectures and more recent transformer architectures (2017).
-
Get an idea of the models and how they are trained
-
Text Embeddings and vector databases, why RAGs are so awesome
-
Advanced prompt engineering methods
-
AutoGen (MS) agent based, Advanced Document QA/multi-modal
- DeepLearningAI Excellent courses!!!