Remote PartTime Data Scientist Multimodal Foundation Model Evaluation & Data

Job Description

```htmlAbout arenaflex Pioneering the Future of Intelligent Computingarenaflex is a global leader in cuttingedge technology, delivering innovative hardware, software, and services that empower millions of users worldwide. Our mission is to blend seamless user experiences with powerful artificial intelligence, creating products that feel intuitive, responsive, and truly personal. As part of our ongoing commitment to push the boundaries of AI research, arenaflex invests heavily in the development of multimodal foundation modelssystems that can understand and generate text, images, video, and moreall within a single unified framework.Why This Role MattersIn todays fastevolving AI landscape, the ability to evaluate, refine, and scale multimodal models is a critical differentiator. arenaflexs Data Quality (DAQ) team is expanding its expertise to include rigorous scientific assessment of these models, ensuring they meet the highest standards of performance, fairness, and reliability. As a Remote PartTime Data Scientist, you will be at the heart of this effort, collaborating with worldclass ML engineers, data analysts, and infrastructure specialists to shape the next generation of intelligent products.Role OverviewThis position blends deep technical research with practical data engineering. You will design and execute evaluation pipelines, develop novel benchmarking methodologies, and contribute to the creation of highquality training datasets. While the role is parttime and fully remote, you will work closely with crossfunctional teams across multiple time zones, participating in regular virtual syncups, code reviews, and design discussions.Key Responsibilities Model Evaluation & Benchmarking: Design, implement, and maintain rigorous evaluation frameworks for largescale multimodal foundation models such as SAM, LLAMA, LLaVA, CGPT4V, and others. Data Pipeline Development: Build robust data ingestion, cleaning, and transformation pipelines that feed highquality data into model training and validation cycles. Statistical Analysis & Reporting: Conduct detailed statistical analyses of model performance, error patterns, and bias metrics; produce clear, actionable reports for engineering and product stakeholders. Experiment Design (DOE): Plan and execute systematic experiments, including ablation studies and largescale user simulations, to uncover insights that drive model improvements. Collaboration & Knowledge Sharing: Partner with ML engineers, data scientists, and infrastructure teams to integrate evaluation tools into the broader ML workflow; mentor junior team members on best practices. Feature Specification & User Impact Modeling: Translate datadriven findings into feature specifications that anticipate user experience outcomes and guide product roadmaps. Tool Development: Create reusable software utilities for data visualization, model diagnostics, and automated reporting using Python and associated scientific libraries.Essential Qualifications Bachelors degree in Computer Science, Statistics, Applied Mathematics, or a related quantitative field. Minimum of 3 years of professional experience in data science, machine learning, or AI research, preferably within a hightech or researchintensive environment. Strong foundation in machine learning theory, computer vision, and deep learning architectures. Demonstrated expertise in evaluating complex AI models, including experience with performance metrics, error analysis, and bias detection. Proficiency in Python programming; comfortable with libraries such as Jupyter, Pandas, NumPy, Matplotlib, and scientific computing tools. Handson experience with deep learning frameworks (e.g., PyTorch, TensorFlow, JAX) for model training and inference. Excellent written and verbal communication skills, with a proven ability to convey technical concepts to diverse audiences.Preferred Qualifications & Additional Skills Masters or Ph.D. in a quantitative discipline, with a focus on AI, computer vision, or multimodal learning. Experience working on largescale foundation models (e.g., SAM, LLAMA, LLaVA, CGPT4V) and familiarity with their architectural nuances. Background in statistical experiment design, hypothesis testing, and causal inference. Knowledge of data annotation pipelines, crowdsourcing platforms, and quality assurance processes for training data. Familiarity with cloudbased ML infrastructure (AWS, GCP, Azure) and containerization technologies (Docker, Kubernetes). Track record of publishing research findings in peerreviewed conferences or journals. Ability to thrive in a remote, parttime setting while maintaining high productivity and meeting project deadlines.Core Skills & Competencies Analytical Rigor: Ability to dissect complex model behaviors, identify root causes of performance gaps, and propose datadriven remediation strategies. Collaboration: Strong teamwork .

Keyskills :
PythonMachine LearningDeep LearningData ScienceStatistical AnalysisData EngineeringBenchmarkingCollaborationModel Evaluation