Machine Learning Engineer

About This Role

This foundational role will shape how Abaka builds, trains, and optimizes multimodal AI systems. You will own the design and development of scalable training pipelines, work directly with data engineering and research teams, and help drive the technical roadmap for model development across multiple modalities. As an early member of the engineering team, you will influence core decisions around model training strategy, experimentation frameworks, distributed infrastructure, and internal best practices.

Responsibilities

• Design, build, and optimize scalable machine learning pipelines for multimodal model training, fine-tuning, and evaluation across text, image, audio, video, and 3D data
• Work closely with data engineering and research teams to develop efficient data workflows, including collection, preprocessing, annotation, versioning, and model integration
• Implement and refine training strategies for large-scale AI systems, including vision, video, and diffusion models, ensuring reproducibility, efficiency, and strong model performance
• Develop tools and automation frameworks that accelerate model experimentation, hyperparameter tuning, and deployment
• Identify and address performance bottlenecks in data or training pipelines to improve throughput, stability, and resource utilization
• Collaborate with product and infrastructure teams to ensure smooth integration of model outputs into both internal and client-facing applications
• Support internal best practices for model governance, experiment tracking, and documentation to maintain high engineering standards and reproducibility

Requirements

• Strong academic background in computer science, artificial intelligence, machine learning, or related fields. Master’s degree or Ph.D. is preferred
• 3+ years of experience in applied machine learning or ML engineering, with a demonstrated ability to deliver production-ready models or pipelines
• Proficient in Python and ML frameworks such as PyTorch, TensorFlow, or JAX, with hands-on experience in large-scale distributed training and inference systems
• Familiarity with multimodal data processing (e.g., text-image pairing, video understanding, speech-audio modeling) and dataset optimization for model training
• Solid understanding of ML system design, including feature pipelines, data loaders, model serving, and evaluation frameworks
• Experience with modern infrastructure tools such as Kubernetes, Ray, Airflow, or MLflow, along with cloud-based training environments (AWS, GCP, Azure)
• Excellent communication and collaboration skills, capable of working effectively across engineering, research, and product teams to accomplish shared goals
• Self-driven and adaptable, comfortable operating in a fast-paced startup environment, and able to demonstrate strong ownership and urgency in execution