.jpg)
Hackathons bring together the brightest minds in the tech world to create something that the world has never seen. That’s why we’re thrilled to co-sponsor the Inference Time Compute Hackathon alongside Anthropic, Cognition, Mercor, and Etched—a 24-hour competition focused on building next-generation reasoning applications using inference-time computing.
This AI hackathon will bring together over 150 AI researchers, engineers and machine learning experts to compete in teams of four, designing and optimizing AI systems. with a shared goal of pushing our industry forward through innovative thinking and execution. Apply now to be part of this exciting hackathon, with a chance to win $60,000 in total prizes.
📅 Event details:
Where: 55 2nd Street, San Francisco, CA, 94105 (Mercor Headquarters)
When: February 28 – March 1, 2025
Hosts: Mercor, Etched and Cognition
Award: $60,000 in total prizes, with $40,000 cash going to the winning team
In the last few months, we have witnessed impressive advancements in this space, with recent lifts from test time compute and reinforcement tuning unlocking new capabilities for reasoning models. This has enabled us to move beyond structured outputs and formatting changes from fine-tuning to real, steerable reasoning level improvements. We are excited to see what participants can do with this technology.
What makes this AI hackathon unique?
Unlike traditional hackathons, this AI hackathon is focused on inference-time compute, an essential component of real-world AI applications. While model training gets most of the spotlight, inference performance is critical for AI models to operate efficiently at scale.
Participants will work on hackathon projects that explore:
🔹 Cutting-edge inference pipelines to drive speed and reduce compute overhead.
🔹 Fine-tuning large-scale models for steerable reasoning and real-time decision-making.
🔹 Developing next-gen AI applications using platforms like Claude (Anthropic), DeepSeek, and Cognition’s coding models.
Teams of four will have 24 hours to build something innovative that incorporates real-time AI reasoning—a fast-growing frontier in AI development.
Powering AI innovation with high-performance compute
Inference-time compute is a hardware-intensive process that requires high-speed, low-latency GPUs to execute complex AI tasks efficiently. As a leader in AI cloud infrastructure, CoreWeave provides participating teams with cutting-edge compute resources to bring their ideas to life.
CoreWeave’s infrastructure is designed to handle demanding AI inference workloads, offering:
⚡ Scalable NVIDIA GPUs to accelerate real-time decision-making.
⚡ Optimized compute clusters to maximize performance and efficiency.
⚡ AI-native cloud infrastructure that supports inference-heavy applications.
By leveraging CoreWeave’s compute power, participants can build next-gen inference models that optimize real-world applications—whether in finance, healthcare, gaming, or automation.
Meet the AI hackathon hosts & sponsors
The Inference Time Compute AI Hackathon is organized by Mercor, Cognition, and Etched, leading companies in reasoning-based AI models and intelligent automation. These organizations specialize in building LLMs and AI-driven tools that push the boundaries of machine intelligence.
CoreWeave is co-sponsoring this event alongside Anthropic, one of the top AI research labs driving the development of Claude and other advanced language models. Together, we are providing the compute, tools, and technical expertise needed to foster AI innovation at this hackathon.
How to get involved in the inference time compute hackathon:
📌 Who Should Apply?
This event is designed for AI researchers, ML engineers, and infrastructure specialists who want to work on groundbreaking AI hackathon projects focused on reasoning models and inference-time compute.
📌 How Do Teams Win?
The winning project will showcase exceptional AI reasoning capabilities, leveraging efficient inference techniques and compute resources to build a functional, scalable application.
📌 How to Apply?
Spots are limited, and applications are being reviewed on a rolling basis. Apply to secure your place today.
