About the job
Amazon's Selling Partner Support handles tens of millions of contacts annually worldwide. The Titans Science team is transforming this experience by building AI agents that autonomously resolve seller issues, learn from every interaction, and continuously improve with minimal human intervention. These agents reason, remember, and adapt — from understanding the seller's context and selecting the right solution, to routing contacts optimally, automating resolution end-to-end, and augmenting associates with AI when human judgment is needed. We do this in deep partnership with multiple engineering and product partners.
Responsibilities
- Own end-to-end research and development of RL-based agent improvement systems — from problem formulation through production deployment and impact measurement.
- Design novel approaches to preference learning, reward modeling, and policy optimization in the context of conversational agents operating over real-world tools and APIs.
- Build and maintain evaluation frameworks that measure agent quality across multiple dimensions: helpfulness, correctness, safety, and alignment with operational standards.
- Collaborate with a team of scientists that work on forefront of Natural Language Understanding, Optimization, Machine Learning and Statistics
- Partner with 10+ engineering teams to deploy models into production systems serving sellers worldwide.
- Publish research at top venues (NeurIPS, ICML, EMNLP, AMLC) — the complexity of our problems produces publishable work, and we actively support it.
- Raise the scientific bar through rigorous peer review, mentorship of junior scientists, and contribution to hiring.
Qualifications
Minimum
- 7+ years of applied research experience
- 5+ years of building machine learning models for business application experience
- PhD, or Master's degree and 5+ years of applied research experience
- Experience programming in Java, C++, Python or related language
- Experience with neural deep learning methods and machine learning
Preferred
- Experience with modeling tools such as R, scikit-learn, Spark MLLib, MxNet, Tensorflow, numpy, scipy etc.
- Experience in a variety of design, wire-framing, and prototyping tools
- Demonstrated experience leveraging generative AI tools to enhance workflow efficiency and productivity, with the ability to craft effective prompts and critically evaluate AI-generated outputs in a professional setting
- Experience identifying opportunities to integrate AI solutions into products and services to drive business value.
- Building and Scaling Agentic System Components like Memory, Retrieval, Reasoning and Tool Calling