CHALLENGES

Challenge 01

Weight : 5

RoboVision Challenge

Develop a model that predicts multi-step pick-and-place sequences for robots from a single perspective camera image. Participants must detect objects, understand spatial relationships, and translate 2D observations into 3D world coordinates to plan optimal manipulation actions.

Go to Kaggle competition

Challenge 02

Weight : 5

QuantumQuest: The Initial Mapping

Build a solution that maps logical qubits to physical qubits on quantum devices to improve program efficiency and reduce errors. This challenge focuses on solving an optimization problem critical for running quantum programs effectively.

Go to Kaggle competition

Challenge 03

Weight : 4

Pharma-Semantic Challenge

Create a deep learning model that reads pharmacological properties of drug pairs and produces human-readable clinical alerts. The goal is to help healthcare professionals manage drug-drug interactions and improve patient safety.

Go to Kaggle competition

Challenge 04

Weight : 3

Deep-Pharma Multimodal Intelligence

Build a multi-task machine learning model to predict drug-drug interactions, including severity, side effects, and population likelihood. Participants aim to prevent adverse events and reduce healthcare costs using predictive AI.

Go to Kaggle competition

Challenge 05

Weight : 2

SeaFloor Pattern Classification Challenge

Create a model to classify patterns on the seafloor from underwater data. Participants explore unmapped areas to detect geological and biological structures, revealing insights about Earth’s underwater landscapes.

Go to Kaggle competition

Challenge 06

Weight : 4

Well Log Reconstruction Challenge

Develop a model to predict missing well log measurements from real LAS files. Participants handle multi-target regression, noisy and incomplete data, and replicate workflows used in petroleum reservoir characterization.

Go to Kaggle competition

Challenge 07

Weight : 2

Gradient Ghostes Challenge

Investigate the catastrophic failure of ORACLE, the flagship trading model at Meridian Capital, and decrypt a 48-byte file left behind by its missing creator. Participants must analyze market data, uncover hidden model blind spots, and extract three failure-based keys

Go to Kaggle competition