Senior Applied Scientist - SSG Science

About Amazon Devices

Amazon Devices leads innovation by designing and engineering prominent technology such as Kindle, Fire Tablets, Fire TV, Health Wellness, Echo, and Astro products. Join this dynamic team to advance cutting-edge techniques for deploying Generative AI models optimized to run efficiently on edge devices through co-design with custom machine learning hardware architectures.

Key Responsibilities

• Optimize Generative AI models for edge deployment via techniques like quantization, pruning, distillation, and fine-tuning.
• Gain deep insights into Amazon's Neural Edge Engine architecture to pioneer novel model optimization approaches.
• Analyze deep learning workloads and advise on their integration with the Neural Edge Engine hardware.
• Apply foundational principles from Information Theory, Scientific Computing, Deep Learning Theory, and Non-Equilibrium Thermodynamics to model development.
• Train state-of-the-art custom Generative AI models exceeding current performance benchmarks to enable future production deployments.
• Collaborate closely with compiler engineers, applied scientists, hardware architects, and product teams to create optimal ML-based solutions for Amazon devices.
• Represent Amazon within the machine learning community by publishing open-source projects and presenting at top conferences such as NeurIPS, ICLR, and MLSys.

Required Qualifications

• At least 3 years experience building machine learning models applied to business challenges.
• PhD or a Master’s degree coupled with minimum 4 years of research experience in applied machine learning.
• Proficiency in programming with Java, C++, Python, or equivalent languages.
• Solid background in neural deep learning techniques and general machine learning methods.
• Demonstrated experience in deploying ML models into production environments, managing the entire lifecycle from development to operational use.

Preferred Qualifications

• Knowledge and ideally hands-on experience optimizing inference workflows using advanced methods such as Mixture-of-Experts and Diffusion Models for language generation.

Additional Information

Amazon is committed to an inclusive culture that empowers employees to deliver excellent outcomes. Candidates requiring accommodations during the hiring process due to disabilities are encouraged to inquire accordingly. The position is based at the ADCI - BLR 14 SEZ location in Bengaluru, Karnataka.