Principal Research Scientist, Amazon Connect

Do you want to define the scientific direction for how a global contact-center network forecasts demand, schedules its workforce, and optimizes operations in real time? The Eliza team within Amazon Connect (FCS) is looking for a Principal Research Scientist with a deep operations research specialization to set the research agenda in operations science — combinatorial optimization, queueing theory, stochastic modeling, and forecasting — and to translate that research into production systems that serve millions of customer interactions. As a Principal Research Scientist, you will be the senior technical voice for operations research across the org. You will identify the highest-leverage scientific problems, architect novel solutions, and drive them from research through production deployment. You will not sit apart from the work — you will remain deeply hands-on with data, models, and systems while raising the scientific bar for scientists and engineers around you. This is an individual-contributor Principal role: your influence comes from technical depth, invention, and the ability to move business and engineering roadmaps through scientific rigor. You will operate at the intersection of demand forecasting, workforce scheduling, and network optimization — applying combinatorial optimization to large-scale scheduling and resource-allocation problems and queueing theory to model contact-center dynamics under uncertainty, non-stationarity, and competing operational constraints at Internet scale — turning that reasoning into systems that continuously sense, predict, and optimize. Key job responsibilities - Set the multi-year research direction for operations research across contact-center demand prediction, workforce scheduling, and network optimization. - Design and deliver novel algorithms in combinatorial optimization (large-scale scheduling, resource allocation, integer/constraint programming) and queueing theory (contact-center modeling, staffing under stochastic arrivals), alongside stochastic modeling and time-series forecasting, that advance the state of the art while solving real operational problems. - Own end-to-end scientific solutions — from problem formulation and prototyping to production deployment — ensuring robustness, explainability, and seamless integration with existing systems. - Design rigorous experiments and evaluation methodology to validate hypotheses and quantify business impact; establish scientific-excellence mechanisms (metrics, benchmarks, peer review) that the broader science team adopts. - Partner with engineering, product, and operations teams to define data and logging requirements, get them prioritized on roadmaps, and translate scientific capabilities into measurable business outcomes. - Influence senior leadership through written papers and deep-dives, framing complex algorithmic trade-offs in clear business terms. - Mentor and raise the bar for applied and research scientists across the org while maintaining significant hands-on technical contribution. - Represent the team's science externally where appropriate (publications, patents, top-tier venues such as INFORMS, NeurIPS, ICML). A day in the life Your day blends hands-on science with technical leadership. You might spend the morning deep in data and models — prototyping a new combinatorial-optimization formulation for workforce scheduling or a queueing model for staffing under stochastic arrivals against production infrastructure — and the afternoon guiding fellow scientists through a hard optimization or stochastic-modeling problem, reviewing an experiment design, or aligning engineering partners on the data architecture needed to unlock the next capability. You'll drive technical discussions with the team and key stakeholders, and periodically write and present papers that shape the business and engineering roadmap.