Publications

A. Ostapenko, S. Wintner, M. Fricke, Y. Tsvetkov. Speaker Information Can Guide Models to Better Inductive Biases: A Case Study On Predicting Code-Switching. ACL 2022 (Accepted).

S. Arora, A. Ostapenko, V. Viswanathan, S. Dalmia, Florian Metze, Shinji Watanabe, Alan W. Black. Rethinking End-to-End Evaluation of Decomposable Tasks: A Case Study on Spoken Language Understanding. Interspeech 2021.

L. Specia et al. Grounded Seuqence to Sequence Transduction. In IEEE J. Sel. Top. Signal Processing 2020.

R. Valente, A. Ostapenko, B. C. Sousa, J. Grubbs, C. J. Massar, D L. Cote, R. Neamtu. Classifying Powder Flowability for Cold Spray Additive Manufacturing Using Machine Learning. IEEE Big Data 2020.

A. Ostapenko, R. Neamtu, F. Anderson: FinDX: A Versatile, Low-Resource Approach to Financial Website Classification. IEEE BigData 2019.

Internships & Teaching

Amazon Robotics – Software Engineer Intern (May 2020 - August 2020)
State Street Corporation – Cognitive Software Engineer Intern (June 2019 - August 2019)
Smartvid.io – Research Assistant Intern (May 2017 - August 2017)
Teaching Assistant (WPI)
        Object-Oriented Design Concepts (Java 8) - Fall 2018
        Linear Algebra II - Spring 2019

Select Experiences

Johns Hopkins University’s Fifth Frederick Jelinek Language Technology Workshop

...

I was in Dr. Lucia Specia’s multimodal machine translation team (MMT). I helped develop deep, context-aware techniques which, given a captioned image, could use the image to resolve ambiguities when translating text. MMT was part of the Grounded Sequence to Sequence Transduction team, a larger effort to understand actions in instructional videos by aligning audio, visual data, and video transcriptions. (May-August 2018)

Relevant Work:
L. Specia, J. Wang, S.J. Lee, A. Ostapenko, P. Madyastha. "Read, Spot and Translate." Springer Machine Translation Journals, (accepted November 2019). To appear.

Applied Machine Learning Research with Vestigo Ventures

From Spring 2019-2020, I worked on applied research projects with Vestigo Ventures, an early-stage venture capital firm. In my first project, I developed a versatile, ML-driven automatic website classification tool which allows Vestigo to identify new investment opportunities from the World Wide Web.

...

Relevant Work:
A. Ostapenko, R. Neamtu, and F. Anderson. "FinDX: A Versatile, Low Resource Approach to Financial Website Classification." IEEE Big Data 2019, Third International Workshop on Big Data for Financial News and Data. To appear.

Oct-Dec 2019: I am completed my senior capstone project at Vestigo Ventures and Cogo Labs. I worked with two other WPI students to build a clickstream data analysis tool which Vestigo & Cogo can use to visualize and analyze how users navigate through company websites. Our tool draws upon concepts in computer science, mathematics, and industrial engineering.