Overview
Want to apply your technical skills and innovative ideas on top of the collective financial data of 40+ million consumers and small businesses, and help build data products (in TurboTax, QuickBooks, and Credit Karma) that solve real-world customer problems and power prosperity around the world?
Areas we are exploring:
- Times series forecasting
- Knowledge Engineering
- Reinforcement Learning/Bandits/Causal Inference
- Image/document understanding
- Intent classification
- NLP/NLU/NLG
- Conversational UI, Chatbots
- Personalization and recommendation
- Deep learning
- Semi-supervised learning
- ML services (autoML, feature recommendation, explainable AI, etc)
What you'll bring
- Must be currently enrolled in a degree seeking program and return to school after internship is complete
- PhD in Computer Science or related technical field (will consider Masters students with previous experience)
- Familiar with machine learning techniques (regression, classification, clustering, optimization, etc) and understand their mathematical foundations
- Ability to explore, discover and import data from multiple sources and make them machine learning ready
- Design and test hypotheses about causes and cures
- Strong programming skills (Python and Scala preferred)
- Excellent communication skills and ability to learn fast
- Experience in developing machine learning solutions to solve real-world problems is a plus
- Experience with Hadoop or Spark is a plus
- Published works in top tier Data Science and machine learning conferences such as KDD, ICML, NIPS, ICLR, ACL, SIGIR, WWW, CVPR, SIGMOD, etc. is a plus
How you will lead
- Adversarial Deep Learning ranking algorithms for question-answer forums
- Use topic modeling to link form lines to verbose instruction/publication documents
- Patch-based information extraction using an unsupervised form segmentation algorithm
- Assess agent call quality using call transcript data
- Apply deep learning for transaction time series forecasting with uncertainty estimation
- Explore active learning to improve the event labels used to train our supervised models
- Predict cognitive biases using financial data
- A Deep-Learning Approach to building Temporal Recommendation Models
- Real-Time Churn Prediction Models
- Developing an unsupervised knowledge acquisition for question answering with Pre-trained Language Models
- Developing a customer intent classifier model for Intuit Digital Assistant