Hi! I'm an M.S. Computer Science and Engineering student at The University of Michigan, Ann Arbor.
I am currently a student researcher at UM's Transportation Research Institute working on robust methods to determine the posture of vehicle occupants.
Prior to joining UM, I was at Microsoft, where I worked on automated techniques to filter out junk content in Bing's search index.
My primary research interests are:
- Trustworthy ML - quantifying uncertainty, model robustness, and the effectiveness of explanation techniques
- Data-Constrained ML - dealing with insufficient and/or noisy data, or the cost associated with annotating data
- ML for Human Safety - creating trustworthy ML solutions for safety-critical environments to reduce human error
Anuj Tambwekar
Contact: anujt@umich.edu
Based on the Plain Academic template
Last updated: 22nd October 2023
Publications
* indicates joint first author- A. Tambwekar, A. Maiya, S. Dhavala, and S. Saha, Estimation and Applications of Quantiles in Deep Binary Classification, IEEE Transactions on Artificial Intelligence [Link]
- A. Tambwekar, K. Agrawal, A. Majee, and A. Subramanian, Few-Shot Batch Incremental Road Object Detection via Detector Fusion, Proceedings of The IEEE/CVF International Conference on Computer Vision (ICCV) Workshops 2021 [Link]
- M. Kashyap*, A. Tambwekar*, K. Manohara, and S. Natarajan, Speech Denoising without Clean Data: A Noise2Noise Approach, Proceedings of Interspeech 2021 [Link]
Experience
For additional details, please check my CV.Research Assistant – University of Michigan Transportation Research Institute
Jan 2023 - Present
- Developing computer vision models to determine vehicle occupant posture using LIDAR and IR data
Engineering Intern – GPTfu Inc.
May 2023 - Aug 2023
- Backend and ML Engineering for an early-stage generative AI startup
Software Engineer – Microsoft
July 2021- Aug 2022
- Automated techniques to reduce junk, spam, and unwanted pages in the Bing search index, while reducing false positives
- Development of a low-latency shared-memory LPC library for same-node interprocess communication.
Research Intern – Intel
Jan 2021 - May 2021
- Created a new architecture for few-shot incremental object detection that achieved SOTA performance on the India Driving Dataset's Open Set task.
Teaching
I've been a Graduate Student Instructor for the following courses at UM- EECS 492: Introduction to Artificial Intelligence (Fall 2023 - 341 students | Fall 2022 - 274 students)
- EECS 448: Human-Centered ML (Winter 2023 - 73 students)