ApertureData Challenge: Build Multimodal AI Applications
ApertureData Challenge: Build Multimodal AI Applications
For questions, please join our Slack channel (preferred for quick responses) or write to team@aperturedata.io
Contest Overview
The world of AI is evolving rapidly, and now is your chance to be at the forefront of this exciting transformation!
Join our ApertureDB contest and experiment with a leading AI database while competing for fantastic prizes. This is a unique opportunity to upgrade your skills while showing off your best work in crafting innovative AI applications powered by ApertureDB.
We have lined up some fantastic judges from AWS, Google, and ApertureData!
In addition to cash prizes, winners will also receive AWS credits and the chance to showcase their work in a "Lunch and Learn" session or through a featured blog post, reaching AI enthusiasts and industry leaders. Ready to Get Started?
The Contest will run from October 10th - Nov 20th, 2024, with submissions due on Nov 20th.
What to Build
There are many creative solutions that you can build using ApertureDB. For any application you develop, you're required to use ApertureDB Cloud.
ApertureDB offers a query interface to manage images, videos, blobs, regions of interest, clips, frames, various application metadata types, and embeddings. It has integrations with PyTorch, Tensorflow, LangChain, and other toolchains.
Below are three examples to get your ideas flowing:
Create a pipeline to extract various modalities of data by using tools like unstructured.io that process and extract metadata / embeddings, and insert all that along with the data in ApertureDB for relevant queries based on the ingested data.
Example: An equipment manufacturer has extensive manuals for their parts and needs a quick way for customer support agents to search for error codes, manufacturer details, and more. These PDF manuals include text, images, tables, and relationships between parts. By automating data parsing and ingesting it into ApertureDB, they can leverage its graph and vector search features to create a question-answering app for agents.Build an application to demonstrate the power of GraphRAG on ApertureDB. GraphRAG enhances RAG responses by utilizing a knowledge graph and vector search. Tools like Whyhow.ai can help you extract the entities and relationships as well as embeddings.
Example: You have doctors' notes, MRI/CT scans, treatment details, and clinical research papers. To assist in diagnosing patients and identifying treatment options, you want to create a graph of this information. By ingesting and parsing these materials into ApertureDB, you can use graphs and embeddings for advanced searches.Personalized recommendation based on visual and other features as well as social media content like testimonials for a large e-commerce vendor. You can source images, videos, anything that users post on social media for any brand and create a knowledge graph of it in ApertureDB to enable a more holistic view for recommending products. You can always index embeddings to do find-similar for any product.
For additional ideas, explore our technical resources or join the ApertureDB Slack channel to connect with experts and fellow developers.
Technical Resources
ApertureData Resources
ApertureDB cloud setup and login instructions
ApertureDB client setup and configuration
Basics of ApertureDB query language
Loading various modalities of datasets
Some example applications
AI/ML Resources
Contest Process
Sign up for a 30-day free trial of your ApertureDB instance with a new cloud account
Refer to our Quick Start Guide
Define your project and start building…
Post on Social Media:
We encourage you to post your 2–3 minute demo video of your multimodal AI project on Twitter, LinkedIn, or Reddit with the hashtags #ApertureDataContest and #ApertureDB and tag ApertureData Linkedin social media account.Submit your entry: Submission Deadline: November 20th, 11:59 PST
How to Submit
Please use this form to submit the following:
5 minute Loom Video that describes the following:
The use case
The challenges you solved
How you incorporated ApertureDB into the workflow you used
A screen share showing the successful execution of the project
Link to your public GitHub repository, with a README, and your code
Participants Pre-Requisites
A few years of industry experience and Python programming are all we expect you need. You can team up but limit team sizes up to 4 people.
Contest Specifications
Objective: Utilize ApertureDB Cloud and any other relevant partner tools that you have access to, to solve for the main challenges focused on AI-driven data management .
Data Types: You’ll work with a variety of multimodal datasets (metadata, embeddings, text, image, video, and so on).
Tools Available: ApertureDB cloud and any third-party integrations that you have access to on GCP or AWS cloud platforms
Winners Announced: December 5th, 2024
Join us on Slack for any questions, suggestions, or the upcoming AMA!
Contest Prizes
We’ve got an exciting prize package for the top 3 winners:
First Prize: $800 per team, plus an ApertureData-sponsored Lunch & Learn to showcase your project*. Alternatively, we can feature your project in our newsletter or promote it through a blog post.
Second Prize: $600 per team and online showcase*
Third Prize: $400 per team and online showcase*
The first 5 submissions receive 2 free months of ApertureDB Basic tier for 1 database instance.
AWS & GCP credits will be announced at a later date.
The prize not only includes the cash incentives above but will also grant visibility in our network of top AI and data leaders.
*With your permission, we will add the winning projects as "Community Project Showcase" on our website and promote them on social channels
Judging Criteria
Judges
Navjot Singh - Data and AI Customer @ Google
Nivedita Kumari - Customer Engineer, Data Analytics @ Google
Suman Debnath - Principal Developer Advocate, Machine Learning @ Amazon Web Services (AWS)
Gautam Saluja - Sr. Software Engineer, Aperturedata
What we are looking for
The contest aims to highlight your data wrangling, querying, and AI skills using multimodal datasets, while leveraging the unique features of ApertureDB. Submissions will be evaluated based on the following criteria:
Use case: Evaluates the project's effectiveness in addressing real-world needs, its level of innovation, use of multimodal data and AI in any given industry, and its user-friendliness for the target audience.
Data wrangling and querying skills: Use of different modalities of data and how you can exercise the multimodal features offered via ApertureDB. Our query language is different as compared to classic SQL or GraphQL options that are popular but specially designed to allow working with property graphs, embeddings, complex data, and ML pipelines. We want to see how well you incorporate the features in building the example use case.
Content: Our primary submission requirement is a video, but we encourage you to include additional materials to help us understand and evaluate your project. Consider submitting blog-ready content, tutorials, or a README file. We’re excited to showcase and amplify the work of our winners!
Dataset: It’s perfectly fine to demonstrate your example with a single data point, but showcasing how it performs with realistic data—whether from public datasets, synthetic sources, or any other accessible data—will strengthen your case. This approach not only validates your claims but also helps you create a more adaptable example for your own use case, provided you have the rights to use that data.
We Welcome Your Feedback:
https://docs.google.com/forms/d/e/1FAIpQLSdhb3JH42fxdvfwNnabbsaW89wAfrRKjQBwEPbIlGWzu3fJ1g/viewform
ApertureData Team