Competitive Data Scientist, Kaggle

• Design challenging data science, machine learning, and mathematical modeling tasks based on real-world problems and competitive programming experience
• Build reproducible execution environments, prepare datasets, configure dependencies, and ensure consistent task execution
• Implement accurate, efficient, and verifiable reference solutions
• Develop automated tests, validation criteria, and evaluation frameworks to distinguish correct solutions from incomplete, inefficient, or flawed implementations
• Run AI coding agents against designed tasks, analyze performance, investigate failure patterns, and refine task specifications
• Identify implementation bugs, incorrect assumptions, data leakage, methodological weaknesses, and other issues affecting solution quality
• Document problem statements, expected outputs, solution methodologies, dependencies, and verification procedures
• Improve task complexity, reproducibility, and evaluation reliability through iterative testing and experimentation

• Demonstrated experience in data science, machine learning, quantitative modeling, applied mathematics, or a related technical field
• Participation in competitive data science or machine learning platforms, such as Kaggle, DrivenData, Zindi, AIcrowd, Numerai, DataSource.ai, or CodaLab/Codabench
• Strong analytical reasoning and ability to break down complex problems into structured, testable solutions
• Practical programming skills sufficient to independently implement, test, debug, and validate technical solutions
• Understanding of statistical methods, predictive modeling, data preprocessing, feature engineering, and model evaluation
• Ability to identify data leakage, overfitting, flawed assumptions, and methodological errors
• Experience evaluating model performance and validating results against reliable benchmarks
• Strong attention to detail and ability to communicate technical reasoning clearly
• Competitive data science
• Kaggle competitions
• Machine learning
• Python programming
• Statistical analysis and mathematical reasoning
• Data modeling and model evaluation
• Testing, debugging, and solution validation
• Experience with DrivenData, Zindi, AIcrowd, Numerai, or similar competition platforms
• Strong Python development skills, including working with unfamiliar codebases
• Experience with pandas, NumPy, scikit-learn, PyTorch, or TensorFlow
• Knowledge of automated testing, reproducible environments, and dependency management
• Experience developing evaluation benchmarks and automated grading systems
• Familiarity with AI coding agents, LLM-based development tools, and agent evaluation
• Experience in quantitative research, optimization, feature engineering, and advanced statistical modeling

• Fully remote work arrangement
• Flexible schedule — choose preferred working hours and days, including weekends
• Contractor schedule of approximately 15 hours per week