NVIDIA and LlamaIndex Developer Contest
Enter for a chance to win cash prizes, an NVIDIA® GeForce RTX™ 4080 SUPER GPU, DLI credits, and more
Contest Overview
Join developers from around the globe in creating innovative large language model (LLM) applications powered by NVIDIA and LLamaIndex technologies. We encourage you to build retrieval-augmented generation (RAG), agentic, or beyond-agentic RAG applications in any domain that interests you and win one of several exciting prizes.
To get started, explore our technical resources and join the NVIDIA Developer Discord channel to connect with NVIDIA and LLamaIndex experts and other developers.
The contest will run from August 27th to November 10th in several countries. See the contest's Terms and Conditions
What to Build
There are endless creative applications that one can build using NVIDIA and LlamaIndex technologies. The following are a few example applications to give you a head start. For any application you develop, you're required to use one of the NVIDIA technologies (NVIDIA NIM™ inference microservices, NeMo™ Guardrails, NeMo Retriever, NeMo Curator, NVIDIA TensorRT-LLM) along with LlamaIndex framework.
Summarization
You can create summarization applications if you have code documentation or bug reports and want to generate actionable insights, or if you need a quick read of technical research papers and articles. You can also develop a local copilot to summarize meeting notes.
Q&A/Chatbots
One popular scenario where Q&A or chatbot apps can be developed is when you need to answer customer queries, assist with product inquiries, and provide order status updates. This workflow can be applied across various industry domains, such as healthcare, education, entertainment, ecommerce, and more.
RAG/Agents
In simple terms, RAG enhances LLM accuracy by integrating facts from external sources, while agents are systems that can act autonomously. You can develop applications such as Arxiv summarizers, email assistants, weather forecasting, and more using either RAG or agentic workflows.
Technical Resources
To get started developing your application, here are some beginner, intermediate, and advanced developer resources for NVIDIA and LlamaIndex technologies. Also, check out the comprehensive guide for more resources.
Beginner
To get started with a beginner application, you can begin by understanding RAG and exploring the following resources:
NVIDIA NIM for embeddings, reranking and more models in LlamaIndex
Intermediate
If you would like to take the next step in creating a RAG application for your own dataset or applying guardrails to the application, start by curating the datasets with NVIDIA NeMo Curator and creating a vector database for your RAG application.
Additionally, you can control the output of your RAG application using NVIDIA NeMo Guardrails and LlamaIndex.
Advanced
If you have RTX systems and would like to develop a local application, start with either ChatRTX, which uses LlamaIndex and TensorRT-LLM, or if you prefer to develop a multimodal RAG application, you can follow the video “Building Multimodal AI RAG for Enhanced Understanding with LlamaIndex, NVIDIA NIM, and Milvus | LLM App Development.”
Explore various data loaders and agent tools, or create your own custom agent tool from here. Whichever path you choose, we encourage you to ensure that the final project meets the technology requirements of the contest.
Discover more examples like this in our github repository
Contest Process
Step 1: Start Now
Register to join the contest and get a head start with our curated technical resources.
Connect with the community of LLM developers and NVIDIA and LlamaIndex technical experts on the NVIDIA Developer Discord channel.
Step 2: Build Your Project*
Set up your development environment and build your project. It must use one or more of the following NVIDIA technologies and LlamaIndex as part of your workflow for an eligible submission:
Step 3: Share on Social
We encourage you to post your 2–3 minute demo video of your generative AI project on Twitter, LinkedIn, or Instagram with the hashtags #NVIDIADevContest and #LlamaIndex and tag one of the NVIDIA (X/ LinkedIn/ Instagram ) and LlamaIndex social media accounts.
Step 4: Submit Your Entry
Once completed, submit your project by accurately filling in all the required fields in the form for it to be considered an eligible submission. We do not permit team entries for the contest.
*To start, if you are curating a dataset for your application, begin with NeMo Curator. Next, develop a RAG application using NIM microservices and LlamaIndex. To ensure your application is trustworthy, consider using NeMo Guardrails. Finally, for optimizing the app's inference performance, use TensorRT-LLM.
Check out the technical resources section for more examples of using NVIDIA and LlamaIndex technologies.
Prizes
A digital certificate of accomplishment will be awarded for all eligible submissions.
The top project will receive a $5,000 cash prize.
The second top project will receive a $3,000 cash prize.
The third top project will receive a $1,000 cash prize.
The projects ranked 4th to 6th will each receive a GeForce RTX 4080 SUPER GPU.
The projects ranked 7th to 11th will each receive a meeting with an NVIDIA engineer to ask any questions, as well as $200 worth of brev.dev compute credits.
The projects ranked 12th to 16th will each receive DLI-instructor-led workshop credits worth $500.
The projects ranked 17th to 126th will each receive $30 worth of self-paced DLI credits.
Judging Criteria
Real-World Novel Application
Assesses how effectively the project addresses real-world applicability, innovation, and how user-friendly it is for its intended audience.
Technology Integration
Evaluates the effectiveness of the developer’s use of NVIDIA's LLM and LlamaIndex technologies in the project, including the quality of documentation and the creativity in integrating a number of these technologies into their workflow.
Office Hours
By attending office hours, you can ask any questions to NVIDIA and LlamaIndex experts to speed up the development of your project. Additionally, by attending, you grant permission to NVIDIA and its affiliates to use and post discussions from the office hours on NVIDIA channels.
September Sessions
October Sessions
FAQs
Here is the link to the list of countries that are open for the contest.
Yes, you can register using your company or student email address. However, team entries for the contest are not allowed.
Team entries are not permitted in the contest, and only the countries mentioned in the terms and conditions are allowed to participate.
It is required to use one or more of the NVIDIA technologies (NVIDIA NIM microservices, NVIDIA NeMo Retriever, NVIDIA NeMo Guardrails, NVIDIA NeMo Curator, NVIDIA TensorRT-LLM) mentioned in the requirements along with LlamaIndex. In addition to these technologies, you can use others.
It is a must to use one or more of the NVIDIA technologies along with LlamaIndex. In addition to these, you can use other services.
Yes, here is the tutorial on how to deploy a self-hosted NIM.
Yes, some of the NVIDIA NIM models support tool calling. Details can be found here.
Any open-source license should be good, such as Apache 2.0, BSD, or MIT.
The submissions will be judged based on their novel real-world applications and effective technology integration. More details can be found here.
We recommend that you provide any keys in the “Instructions to Run the Project” field on the submission form.