Applied Scientist

Our team in Microsoft Azure AI is working on cutting-edge NLP and Deep Learning models to power Microsoft AI products and build it as a service to serve many external applications.

The team is focused on applying deep learning techniques to solve AI problems such as Natural Language to Code, adaptation of large-scale language models, Question Answering on structured data, including the No-code low-code AI models in Power Platform, the copilot project in the collaboration with Github, question answering on structure/unstructured data, GPT-3 fine-tuning, and various customer-facing products.

In addition, we work on customization and pretraining of large-scale language models for AI services.

We are a hybrid team at the intersection of research and product, with positions ranging from researchers and applied scientists to software engineers.

Our team has the flexibility to tackle challenging and exciting research problems that are highly related to product requirements.

For this position, we are hiring for the NL to Code project to work on no-code AI model for Power Platform and other exciting NL2Code and Safety projects to improve its model quality.

Publication in top conferences as a byproduct is encouraged with lots of opportunities to collaborate with professors and students at top universities, as well as mentoring summer interns.

We track and work on state-of-the-art natural language understanding (NLU) models targeting high-visibility and world-class AI leaderboards.

We have previously held the top positions in a few of them, such #1 in WikiSQL , SciTail , SWAG , SNLI , SQuAD, and the significant human parity achievements in both GLUE and SuperGLUE.

You may also see some of our recent papers such as LoRA , DeBERTa and R-Adam .

We are looking for a passionate, creative, analytical individual who loves NLP, deep learning, machine learning, and wants to ship products quickly at a massive scale.