Hosting Kevinđș â The Worldâs first AI manager in Lightning AI
In a previous post, I talked about creating Kevin, the worldâs first AI manager by giving Mistral-7B-Instruct-v0.2 a personality.
The personality of a manager that âhas your best interests at heartâ and always âwatched out for youâ.
If you didnât get it yet, I am being sarcastic. Kevin is far from friendly.
Hereâs a link to the article -
In continuation of that article, I will now enumerate how I went about deploying Kevin on Lightning AI, thus allowing a limited access to folks to chat with Kevin.
Why Lightning AI?
I was toying with the idea of deploying Kevin on AWS or GCP, but then I saw a few posts about Lightning AI on LinkedIn and decided to check it out.
As you know, I love exploring new tools and technologies
I registered with my email id and once my account was activated, I was given 15 credits to play around with and deploy to a single GPU instance â
(A T4 single GPU instance with 16 GB Memory and 8 vCPUS).
And as you can see, the cost is pretty reasonable.
Also, if you create a Studio on Lightning AI, you can make it public and people can check out your work in all its glory.
NOTE â Kevin isnât public yet and wonât be for the time being
I loved the interface, found it cheaper than most other cloud vendors out there and decided to move ahead with hosting Kevin in Lightning AI.
So how did I go about it?
Deploying to Lightning AI
Once you have registered on Lightning AI and have been re-directed to your dashboard, click the button
To kickstart the process to start a new Studio.
I selected Code â
And clicked on Start.
I got options to choose from IDE like interfaces and Chose Visual Studio Code.
This is what it looks like â
Looks lovely, doesnât it?
Now, on the right hand side you will see a menu and a big âPlusâ sign at the bottom. This is where Lightning AI really shines through.
The Plus sign allows you to add various plugins to create Interfaces/Web Apps etc.
This is what clicking the plus sign shows me â
I needed to create a Web App to expose Kevinâs backend interface so went to Web Apps and chose Streamlit â
That renders the Streamlit plugin along with the other menu items on the right â
Now, when I click on the Streamlit icon, it allows me to run the Streamlit code that I have written.
In my case, this is what the UI looks like when writing the code for Streamlit â
And you can easily run this using the Streamlit icon in the right â
Running the Streamlit app creates a playground instance with a temporary external URL to allow chatting with the bot.
Of course, if you want to host this, you need to reserve an IP and a domain and then point it to the reserved IP in Lightning AI.
Starting the Streamlit app shows me this interface â
You can click on the public link button to get a public URL to access the app.
The logs can show you any issues that you need to fix.
Pretty simple eh?
Other Features in Lightning AI
- Create and Share your Studios
2. Select templates with lots of rich information, guides, code etc to create your own studios
3. Access to educational Blogs, Papers and Tutorials to do stuff in Lighting AI
4. Data Processing, Training and Service services
5. Access to various Models
So obviously, there is so much that you can get done within Lightning AI with easy to use Interfaces and access to information on how to get it done.
Plus the pricing is pretty reasonable as well. Whatâs not to like?
Follow me Ritesh Shergill
for more articles on
đšâđ» Tech
đ©âđ Career advice
đČ User Experience
đ Leadership
I also do
â Career Guidance counselling â https://topmate.io/ritesh_shergill/149890
â Mentor Startups as a Fractional CTO â https://topmate.io/ritesh_shergill/193786