[Music] good afternoon everyone my name is Mohan Yellapantula.
I lead the solutions go to market for connected Mobility solution area with AWS Automotive today
we'll uh I'm here with uh Dr. Humza Akhtar yeah I'm the principle for Manufacturing and Automotive
at MongoDB and today we're going to show you some exciting stuff that we are doing together overview
of our agenda here I'm going to L talk through the AWS Automotive solution areas and then we'll talk
about how do we unlock the full lif
e cycle value of data connected vehicle data and then we'll
talk about gen use cases for connected Mobility that we have worked with mongodb and then also
go in little bit detail about the mongodb atlas developer data platform and then finally how
AWS and MongoDB have come together to Showcase connected vehicle experiences unlocking the full
life cycle of uh a life cycle value of data so a quick overview of the AWS Automotive solution
areas as you can see we have about eight different solut
ion areas and these help address the customer
pain points in different business areas and the key aspect of it is I want to zoom in on connected
mobility and how do you create new experiences and new value generating connected Mobility service is
a key aspect of here and the mongodb actas vector Search tool and AWS gen tools coming together to
Showcase these new experiences is what we're going to uh talk about let me zoom in a little bit more
about the connected Mobility solution area what
are the workloads here what are the use cases
here as you can see we have five use cases here connected vehicle platforms analytics insights
vehicle data collection and also providing uh connected Edge and infrastructure aspects of it
so in this case this use case would fall into the analytics and insights part of it so if you take
the life cycle value of data I want to talk about three personas here who are going to be the end
benefactors of of the data that is coming in from the vehicles
first are the consumers how can they
get new connected experiences during the ownership of of their vehicle so that their experience
is continuously fresh throughout the ownership experience the second is the oems who can take the
uh data that is coming in from these vehicles and then refine the the features inside the vehicle
use it for R&D aspects of it vehicle development aspects of it and three are the actual Mobility
Services providers and partners who can take advantage of the data th
at is coming in from these
vehicles and then keep generating new Mobility Services out of it so as you can look at it it's
both of a internal monetization and an external monetization of the of the data so this part of
it we will take all these three personas here and showcase some use cases one across predictive
maintenance second across building intelligent incar experiences third around route planning and
optimization and Hamza is going to go in detail about how these use cases come toge
ther with
mongod and AWS partners ship all right so let's go about building some use cases now um before
I go into it I just want to explain uh what we offer from mongodb side so on the left hand side
you see the self-managed mongodb database instance uh a lot of people know us as that open source
document model database which is very flexible very fast it powers your real-time applications
but on the right hand side is our fully managed mongodb Atlas developer data platform which
gives yo
u the same database experience as you will get with a on-premise version of MongoDB
but it gives you much much more so uh I'm not going to go through all the features we offer
you can take a photo of this but for connected vehicles and autonomous uh Mobility space we're
going to focus on three aspects here we're going to focus on the database we're going to focus on
Atlas search which is a full teex search engine uh and it also includes Vector search as well
so Powers your gen applications
for connected mobility and we're going to focus on device sync
and device stic which is a little database that goes right inside the car it provides local
storage for the Teltry data that the car is producing and it syncs the data back into the
cloud it will provide you with the bidirectional synchronization uh capabilities right out of the
box so taking these three um offerings that we have let's build a connected vehicle we have a
vehicle and what we're going to do is when to put a device
SDK inside that vehicle so this is a
uh it comes with various sdks there's a C++ SDK K just that's generally available uh we have our
clients actually using the C++ SDK in the car right now so you put that in and this will provide
you with this local object-oriented storage inside the vehicle and then what it will do is it will
use the device sync that is part of MongoDB Atlas offering and you can set up MongoDB atas on AWS
cloud and using device sync it will sync all the data from vehicle
to MongoDB sitting on AWS cloud and
this synchronization is bidirectional it will hand hand all the conflicts for you it will handle
all the network interruptions for you if I'm at a place where there's no internet connectivity
my car will keep storing data locally and then it will synchronize the data to the cloud when
the internet comes back now I can do the same I can put the device SDK in the mobile apps as
well we provide Swift Kotlin Flutter Java SDKs and then you can have the same r
oute established
from data coming in from the car and going into the cloud and then notifications going back into
the mobile app so that is what our clients our OEM customers are using to build these meaningful
uh customer experiences and then you can take advantage of our database triggers and push data
to Amazon event bridge and from there you can set up your Sagemaker model inside Sagemaker Studio
you can run all sort of Predictive Analytics on the data collected from the vehicle send ba
ck
the results into MongoDB send down the alerts using either device sync or you can set up a graph
to endpoint inside MongoDB Atlas and push it down to web applications or any applications of your
choice I will put a lot of QR codes in my slides you can scan this QR code it will take you to
a GitHub Rebo where you can actually set all of this up yourself there step by-step instructions
provided all right so this is how we set up a use case of let's say predictive maintenance for
connected
Vehicles so we provide everything out of the box you just take the building blocks
and set it up so we have customers like Volvo for example Volvo uh uh group connected SLO shells
that is using MongoDB Atlas on AWS to do Fleet Management they're tracking 65 million daily
events coming from all sorts of different trucks out in the road and you can read more about this
story uh by scanning this QR code just going to that link the power that mongodb provides is that
flexibility is that high s
peed availability High availability scalability and then on top of
that it is running on AWS which provides that infrastructure uh that that powers this this
massive application and they can store a lot of telemetry data in their system moving on we we
we talked a little bit about connected Vehicles now I'm going to bring in generative AI aspect
into it so we have we are offering Atlas Vector search as part of our MongoDB offering you can set
it up inside MongoDB Atlas UI uh once you set up
on on AWS Cloud uh and then what what this offers
is actually a semantic search experience so as I mentioned before mongodb is a very very flexible
document database you can store your vector embeddings that are required to power generative
AI applications right next to your data in the database and then once you have your data you can
actually query the database run semantic search queries on top of it so that will remove that
operational heavy lifting you don't have to set up multiple Ve
ctor databases and your operational
databases you do everything in one database how does it relate to AWS Sagmaker jump start for
example there are four components always needed to build any gen AI application you need to have an
embedding model so if you have images if you have technical manuals let's say to fix a car you need
to go through a PDF file for technical manual you have those things you have to embed them you have
to create Vector embeddings out of them you can use uh sagemaker
end point embedding models it
it gives you all the all the access to all the different models you can use one of those to embed
create Vector embeddings store those Vector embeds inside Mongo Atlas and then you can set up a
sage maker end point for llm to ask questions or ask queries through something like Amazon Lex what
mongodb is providing here is the vector search capability it will create that additional context
that is required for your llm model to generate an output that is not hall
ucination you need to
train your llm model on real data you don't want it to hallucinate and give you something wrong
so that uh it it results in a very bad customer experience so that addition context with semantic
search is provided by MongoDB Atlas and then all the embedding models can be enabled via something
like Amazon Sage maker jump start so that's how we work together we published a blog post on it
it gives you step-by-step instructions on how to set this up so take our take take a
moment to
scan this QR code uh and then I will show you how to put this into action for a connected
vehicle use case pretty interesting isn't it so let's go back to our original diagram again
we have the vehicle again we have Atlas device SDK which is a little database that runs inside
the car it's going to transmit the Telemetry data automatically to mongodb setting on AWS cloud
and then we are also connected by the same link to a mobile application now we can run Telemetry
based Diagnos
tics easily I showed you this before but now let's let's put in the vector search and
generative AI capability into it so now what you can do is you can have audio files generated from
the car you can actually record the sound that your car is making your car engine is making for
for ice cars and and then create Vector embeddings on it using Sage models um uh CED for you through
Sage maker and push those Vector embeddings in the D same database which is MongoDB after that you
can set up yo
ur llm model uh using Sagemaker B and connect that with MongoDB Vector search so
using MongoDB Vector search it will give the additional context to the llm model so that it
can generate the right response to your query in the same database you have the Telemetry data you
have the audio files Vector embeddings you have the technical manual embeddings inside and then
you have a predictive maintenance model running in sagemaker studio so we're running all sorts
of analysis here now when you t
ake your car to a technician for after sale support they can use
something like Amazon Lex and do this audio based Diagnostics they can ask a question that this
car is generating this kind of knocking sound and tell me exactly what I need to do to fix it and
what Vector search will do is it will perform a semantic search on those audio embeddings and it
will perform a semantic search on the technical manual embeddings and it will give the right
context to the llm model so that it can genera
te the right response a very concise response a non
hallucination response back to the technician so that they can easily uh go about uh their work
and fix the car and that will result in a very delightful customer experience in my opinion so
that's how we work together that's how you can set up uh you can leverage uh Atlas Vector Search
you can leverage AWS for root CA diagnostics for a connected vehicle and you can ALS set it up
for free um you can uh at least on the MongoDB side we provi
de a free forever cluster and then
you can set up AWS Services as well and uh get [Music] going
Comments