“It’s All About the Long Term” is one of the key themes in the very first shareholder letter of Amazon (https://s2.q4cdn.com/299287126/files/doc_financials/annual/Shareholderletter97.pdf). Long term thinking and decision making , along with customer obsession, has been in Amazon’s DNA from the very beginning. Customer Forecasting & Valuation owns the primary decision metrics at Amazon, leveraging causal machine-learning models to predict the long-term (downstream) impact of customer actions and business decision. These metrics form the foundation of decision taken by an exponentially increasing number of decision makers, both humans and systems, on a day to day basis.

Inferences produced by our systems are used both in real-time systems (such as ads bidders and content recommendation systems) as well as in financial planning activities. In order to serve the wide variety of and rapidly evolving decision contexts throughout Amazon, our models have to work with an ever-increasing volume and complexity of data and at the same time provide inferences with reliability and low-latency.

Engineering robust large-scale Machine Learning Systems presents some unique challenges such as reproducibility and explainability of results, agility and experimentation to evolve rapidly, no-loss translation from experiments to production. We in Customer Forecasting & Valuation are working to tackle these problems as we evolve the next generation of Causal Machine Learning Systems.
You will work in a multidisciplinary team, alongside Research Scientists/Economists, Software engineers and Data-Engineers to build causal machine learning systems processing Several Terabytes of customer-grain data on a daily basis with strict SLAs to influence how billions of dollars at Amazon get spent. You will use a wide-variety of AWS technologies such as AWS Batch, S3, EC2, EMR, DynamoDb along with Spark, Docker and some Amazon internal frameworks and tools.

We are looking for an extraordinary individual who combines superb technical, communication, and analytical capabilities with a demonstrated ability to get the right things done quickly and effectively. We would expect the program manager to understand, question and influence the scientists while raising the bar on engineering machine learning systems.

In this strategic, high-visibility role you will own the execution of cross org projects and customer requests. You will a) work with product managers, science and engineering team within and outside the org to define the scope, execution planning and program updates b) Gather requirements and distill it into clear working-backwards business requirements, c) using broad context from multiple businesses to set and drive 3-year vision and 1-year operating plans, c) set and manage company-wide stakeholder expectations, d) guide a team of 20-25 scientists/economists/software development engineers using clear customer requirements, and e) manage the program aspects including reviews and education sessions.

The product is technically complex. To be successful in this role, you need to be analytical and technical enough to know the science and engineering systems to independently make informed decisions in discussions with customers (and not overly rely on the team in each discussion). You also need the power to generate business insights and guide Directors/VPs with their decision-making. You should enjoy and excel at verbal and written communication, as it will be your job to represent the value of new solutions to business leaders.