Episode 2: The LLM / GPT / AI Prompt / Data Engineer Roadmap

The LLM / GPT Prompt
Engineering Roadmap
Take a peek at what’s to come if you decide to become a
prompt engineer, whether for yourself, your business, or to
work at your next job.
Rahul Xavier Singh Anant Corporation
Nocode Data & AI

Prompt Engineering LLMs
like GPT-3.5/4, Cohere,
Claude, StableLM is like
learning how to “cast magic
spells” More art than
engineering.

Series Agenda
● Introduction to Nocode,
Data, and AI
● Saving Time & Money with
GPT Automation
● GPT Automation: What it is
and How it Works
● Data Analysis Made Easy
with GPT Automation
● Thinking about the Big
Picture

Today’s Agenda
Source: Go-No-Code
1. Levels 1-2 - Everyone
2. Levels 3-4 - NoCode
3. Levels 5-6 - Low Code
4. Levels 7-8 - Software
Engineering
5. Levels 9-10 - Data Engineering
6. Levels 11-12 - ML / Data Scientist

LLM1 & LLM2 : Everyone
● No previous coding experience required.
● Just need to know how to create an
account.
● Might need a credit card …

LLM1: Get Started with Existing ChatBots
1. Start out with ChatGPT, Bing Chat, Bard, or Poe.
2. Learn about the different ChatGPT/GPT
powered apps
3. Choose an app that interests you and start
using it.
4. Experiment with the different features of the
app and see what you can do with it. You can
look into built in automations or plugins
5. Read the documentation for the app and learn
as much as you can about how it works.

LLM2: Become a Prompt Engineer
1. Learn how to prompt on ChatGPT (ShareGPT,
Poe, Github)
2. Experiment with contexts, instructions,
3. Join different Discord groups where people
discuss how to get better at prompt
engineering.
4. Start writing your own prompts, share them
and see if they are useful for others.

LLM3 & LLM4 : NoCode
Prompt Engineering
● Previous coding knowledge helps.
● Should diagram what you need to do.
● Give yourself time to understand the tools.

LLM3: Use GPT API with NoCode
Automation, App Builders
1. Explore NoCode app builders that are available, such
as Bubble, Retool, Softr
2. Learn to use NoCode DBs like Airtable, and Google
Sheets.
3. Get started with NoCode app builder and start using it.
4. Experiment with app builder you chose and see what
you can create with it.
5. Continue to enhance your Platform with plugins and
other tools.

LLM4: Create Workflows to Automate
Tasks with NoCode
1. Learn about NoCode tools that are available for
Automation, such as Zapier, Make
2. Choose a NoCode automation tool with a database
and start using it.
a. Consider using Make/Zapier with Airtable or
Google Sheets.
3. Experiment with the NoCode automation tool and see
what you can do with it.
4. - Continue learning how to enhance Make with other
connectors and tools.

LLM5 & LLM6 : Lowcode
Prompt Engineering
● Previous coding knowledge needed.
● Need to be comfortable with VSCode or
similar.
● Need to know things like REST/HTTP, etc.

LLM5: Use GPT API with Code, make
your own APIs
1. Get started by learning about programming.
a. Easiest to Start: Python, TypeScript/Javascript
(Recommended)
b. Experience Required: Java, Scala, and C#.
2. Experiment with your code using the GPT API. Use GPT
to help guide you.
3. Read the GPT API Docs and learn about different
models.

Tasks with Code
1. Use a Python/Typescript to read from and write to a
database using code.
2. Experiment with different workflows to see how they
can save you time and effort.
3. Share your workflows with others to help them
automate their tasks.
4. For example, you can use Python code with GPT API
and Airtable to automate tasks.
5. ChatGPT can help you with this.

LLM7 & LLM8 : Software /
Prompt Engineering
● Previous software development experience.
● Need to be comfortable debugging, etc.
● Should understand design patterns, UI, API,
Databases, SQL, etc.

LLM7: Use GPT API with your Data / a
Framework
1. Explore how people are working with LLMs with
Tools/Agents using:
a. Llamaindex and Langchain in Python
b. Window.AI and Langchain.js with
JavaScript/TypeScript.
2. Try using GPT API with Langchain and LLAMA Index to:
a. Search Wikipedia/Google
b. Connect to your data via SQLChain, etc.
3. Read documentation and watch videos, join the
LLAMA Index/Langchain Discords.

LLM8: Use GPT API with your Data / a
Framework to Make your own APIs
1. Use the Langchain / LLAMA Index to create your
own APIs that can be used by others.
2. Experiment with ingesting data and create LLM
powered APIs and model real world scenarios.
3. Try out Different VectorDBs, Chunking to see what
works better.
4. Share your APIs with others and help them solve
their problems.

LLM9 & LLM10 : Data /
Prompt Engineering
● Previous data engineering experience.
● Need to be comfortable with batch
processing, parallel processing.
● Should recognize data platforms such as
Databases, Indexes, Vector Databases

Tasks with your Data /a Framework
1. Use the Langchain/LLAMA Index with your data or a
framework to create workflows that automate tasks.
2. Consider using tools like Airflow or DAGster to run
long, complicated workflows.
3. Experiment with vector databases, chunking lengths,
and embedding methods to optimize your workflows.
4. Finally, share your workflows with others and help
them automate their tasks.
5. Continue learning ways to optimize your ingestion /
retrieval / processing engines.

LLM10: Use Another Public LLM API
other than GPT
1. Check out other LLMs like Claude, Cohere,
StableLM, Dolly, and LLAMA.
2. Choose a Python-compatible LLM and use
Langchain to switch between them.
3. Experiment with the different features of the
LLM.
4. Read the LLM documentation to learn more
about how it works.

LLM11 & LLM12 : Deep
Learning / Data Science
● Previous experience in Data Science and
Neural Networks, esp. Transformers
● Need to have your own servers, with GPUs.
● Should be comfortable waiting hours/days to
wait for model training.

LLM11: Fine tune / use open models
on own servers for inference
1. Explore how to refine models and utilize tools such as
Alpaca, Vicuna, and GPT4all.
2. Experiment with refined models and compare with
GPT, Claude, or Cohere.
3. Refine your model using your own dataset. You can
use OpenAI Generated Data for a shortcut.
4. Use your model to do what you were previously
completing with GPT.
5. Have actual users test your model and provide
feedback.

LLM12: Build your own LLM Model
with your own Training Dataset
1. Learn how to make your own LLM model using NN/Deep
Learning
2. Choose a framework you're comfortable with, and start
building your model. Good starting points, papers.
3. Try different parameters to see how they affect your model's
output.
4. Train your model on a dataset of your choice. You’ll need a lot.
(Pile, C4)
5. Use your model to do what you were doing with GPT, like
writing text and translating languages.
6. Ask real users to test it and give you feedback.

24
Key Takeaways for Prompt Engineering Roadmap
Don’t reinvent the wheel.
Let’s do this step by step.
Work with your strengths, and
the strength of others.
Know where you are.
- There are lots of great starting points,
you can get started by looking at how
others are doing it and get started
today.
- You don’t have to train your model on
day one. You probably will never need
to.
- These levels are supposed to be baby
steps that turn into giant steps. Each
level builds on the previous one.
- You need to map your goals in a picture
whether its paper, whiteboard, etc.
- The machine, other projects, other
people can augment your skills.
Map out what you want to do

25
Thank you and Dream Big.
Hire us
- Design Workshops
- Innovation Sprints
- Service Catalog
Anant.us
- Read our Playbook
- Join our Mailing List
- Read up on Data Platforms
- Watch our Videos
- Download Examples

Episode 2: The LLM / GPT / AI Prompt / Data Engineer Roadmap

Recommended

Recommended

More Related Content

What's hot

What's hot (20)

Similar to Episode 2: The LLM / GPT / AI Prompt / Data Engineer Roadmap

Similar to Episode 2: The LLM / GPT / AI Prompt / Data Engineer Roadmap (20)

More from Anant Corporation

More from Anant Corporation (20)

Recently uploaded

Recently uploaded (20)

Episode 2: The LLM / GPT / AI Prompt / Data Engineer Roadmap

Editor's Notes