RNDM:

At RNDM we are building AI agents to improve liquidity in Perp exchanges, Prediction markets. We aim to fundamentally disrupt DeFi with agents. Our agent Atlas deployed on perps generated about 170M in volume this year, and we are in the process scaling with several more agents including lending, LSTs to address all DeFi primitives. We have partnered with several chains/ecosystems including Movement Labs, Flow, Hyperliquidx, Avalanche among others. Our Atlas agent is deployed on Enclave markets in Avalanche, Hyperliquidx and other exchanges. Our Ganesh agent will be on Hyperliquidx, and Apollo will be on Flow L1.

Hackathon scope:

Currently we are using our custom infrastructure to build the agents instead of standard pipelines like LangGraph, and we are more at level 2. Our goal with the hackathon is to explore potential areas where the participants in the hackathon can contribute either by backtesting a strategy with a full simulation resulting in a set of parameters with workflow for use by the agents or building a full scale agent end to end for a DeFi primitive.

Bounty 1: Quant track

You will be provided candle dataset for a few assets (mix of majors,memes), interest rate data. This will be pre-filtered data over a few days, and provided on the first day of the hackathon. Developers have to use existing LLMs/news sources but depending on the use case we can support by guiding to the correct endpoints.

Developers are also free to collect data from lending markets, LSTs, orderbook data from CEXes like Binance for example. The requirement is that any strategy built for an agent atleast has 10k unique datapoints to ensure a meaningful sample size

Goal:

1. Provide a detailed notebook which contains all the steps from data aggregation, strategy design/indicator selection to backtested results.

2. Provide a pseudocode for deploying an agent which will interact with the DeFi primitive, LLMs using the strategy and also formulate all the steps in the agentic workflow.

Prizes: 2k USD in total rewards for 2 submissions.

Bounty 2: Agent track

Your goal is to build an off-chain agent workflow across new DeFi primitives. Please note that preference will be given for a working mvp than a partial product. You can choose to build an agent using the guidelines below. Feel free to build for example your own version of our agents Atlas, Ganesh or Apollo or try one of the following categories below:

1. LSTs,Restaking portfolio agent: Restaking is a significant trend led by Eigenlayer. Our goal as a restaking agent is to navigate the complex landscape of restaking protocols and their payoffs. Use social data, analytics and estimate the dollar value of payoffs, use this to constantly enter and exit a position maximising the staking rewards.

2. Lending agent --> A lending agent is similar to a morpho optimiser The goal is to use an agent to maximise the supply rates using the p2p model, with a twist of using agents to optimise the spread.

Goal:

1. Provide a complete workflow for the agent: interact or use a core model, connect to api for the agent to identify state, execute planning actions and demonstrate the workflow in a short video. Submissions must include working code

2. Explain the methodology i.e. basic backtesting for the strategy, developers can also combine it with bounty 1 to provide a more comprehensive result. Submission in the form of a short paper with backtesting results.

3. Bonus points for using multiple agents such as long+short term inputs to share information, achieve goals.

   Prizes: 8k USD in total with 5k for the best agent, and 3k for the next best submission.

Follow us on X: https://x.com/rndm_IO for announcements

RNDM:

At RNDM we are building AI agents to improve liquidity in Perp exchanges, Prediction markets. We aim to fundamentally disrupt DeFi with agents. Our agent Atlas deployed on perps generated about 170M in volume this year, and we are in the process scaling with several more agents including lending, LSTs to address all DeFi primitives. We have partnered with several chains/ecosystems including Movement Labs, Flow, Hyperliquidx, Avalanche among others. Our Atlas agent is deployed on Enclave markets in Avalanche, Hyperliquidx and other exchanges. Our Ganesh agent will be on Hyperliquidx, and Apollo will be on Flow L1.

Hackathon scope:

Currently we are using our custom infrastructure to build the agents instead of standard pipelines like LangGraph, and we are more at level 2. Our goal with the hackathon is to explore potential areas where the participants in the hackathon can contribute either by backtesting a strategy with a full simulation resulting in a set of parameters with workflow for use by the agents or building a full scale agent end to end for a DeFi primitive.

Bounty 1: Quant track

You will be provided candle dataset for a few assets (mix of majors,memes), interest rate data. This will be pre-filtered data over a few days, and provided on the first day of the hackathon. Developers have to use existing LLMs/news sources but depending on the use case we can support by guiding to the correct endpoints.

Developers are also free to collect data from lending markets, LSTs, orderbook data from CEXes like Binance for example. The requirement is that any strategy built for an agent atleast has 10k unique datapoints to ensure a meaningful sample size

Goal:

1. Provide a detailed notebook which contains all the steps from data aggregation, strategy design/indicator selection to backtested results.

2. Provide a pseudocode for deploying an agent which will interact with the DeFi primitive, LLMs using the strategy and also formulate all the steps in the agentic workflow.

Prizes: 2k USD in total rewards for 2 submissions.

Bounty 2: Agent track

Your goal is to build an off-chain agent workflow across new DeFi primitives. Please note that preference will be given for a working mvp than a partial product. You can choose to build an agent using the guidelines below. Feel free to build for example your own version of our agents Atlas, Ganesh or Apollo or try one of the following categories below:

1. LSTs,Restaking portfolio agent: Restaking is a significant trend led by Eigenlayer. Our goal as a restaking agent is to navigate the complex landscape of restaking protocols and their payoffs. Use social data, analytics and estimate the dollar value of payoffs, use this to constantly enter and exit a position maximising the staking rewards.

2. Lending agent --> A lending agent is similar to a morpho optimiser The goal is to use an agent to maximise the supply rates using the p2p model, with a twist of using agents to optimise the spread.

Goal:

1. Provide a complete workflow for the agent: interact or use a core model, connect to api for the agent to identify state, execute planning actions and demonstrate the workflow in a short video. Submissions must include working code

2. Explain the methodology i.e. basic backtesting for the strategy, developers can also combine it with bounty 1 to provide a more comprehensive result. Submission in the form of a short paper with backtesting results.

3. Bonus points for using multiple agents such as long+short term inputs to share information, achieve goals.

   Prizes: 8k USD in total with 5k for the best agent, and 3k for the next best submission.

Follow us on X: https://x.com/rndm_IO for announcements