AI Developer Intern

Looking for that internship? The game-changing one that’ll help you learn, grow, and chart your path forward? You’ll find it at SAS. Our interns aren’t coffee runners – they do real, meaningful work. Our award-winning internship program is focused on development, culture, and community. We’ll help you grow professionally, find (or further) your passion, and make memorable connections that last beyond the summer.

The Analytics Commons team is responsible for supporting the Analytics R&D division with infrastructure and frameworks for automated machine learning, parallel model tuning, portable model scoring, optimization solvers, core SAS functions, and other predictive modeling utilities. SAS products collectively support a great variety of analytical modeling techniques - including statistical models, machine learning models including deep learning, natural language processing models, forecasting models, recommender models, as well as econometric and time series models. AUTOTUNE is a proprietary framework that facilitates parallel model tuning to drive model hyperparameters towards values that optimize a predictive model to a particular data set. The tuning process can be expensive, but can be aided by previous tuning efforts to similar data.

In this summer internship, you will investigate the ability to extract and log metadata from model tuning jobs towards building a database of model configurations for particular data sets. You will instrument existing C code to retrieve this metadata and build an example metadata database from a variety of example data sets, and you will demonstrate the ability to use this metadata to jumpstart the tuning process. You will incorporate agentic AI into the process of generating SAS code for tuning a predictive model that uses available metadata.

As an intern, you might:

• Investigate Open Telemetry (OTEL) logging from C code

• Instrument AUTOTUNE code to log metadata about training data and hyperparameter configurations while tuning

• Demonstrate the ability to use this metadata to jump start the tuning process

• Expand an AUTOTUNE AI agent to initialize tuning from a metadata library