Achyudh Ram

Hi! I am Achyudh Ram, a research master's student at the University of Waterloo working with the Data Systems Group and the Software Analytics Group. I have previously interned at Microsoft Research, Intuit, and TU Delft; more on my resume.

I am interested in building intelligent systems that aid in the software development process by bringing together diverse fields like natural language processing and software analytics. I write about some of the interesting aspects of my projects in my blog.


Rethinking Complex Neural Network Architectures for Document Classification

Annual Conference of the North American Chapter of the Association for Computational Linguistics (2019)

What Makes a Code Change Easier to Review? An Empirical Investigation on Code Change Reviewability

26th ACM Joint European Software Engineering Conference and Symposium on the Foundations of Software Engineering (2018)

Investigating Type Declaration Mismatches in Python

IEEE Workshop on Machine Learning Techniques for Software Quality Evaluation (2018)

Bidirectional Encoder Representations from Transformers for Document Classification

Preprint arXiv:1904.08398

Supervised Sentiment Classification with CNNs for Diverse SE Datasets

Preprint arXiv:1812.09653

Recent projects


Deep learning models for document classification and relevance transfer implemented in PyTorch


A PyTorch-based framework for building deep learning models on source code


A framework for running ensembles of Seq2Seq Neural Machine Translation models on PySpark using Elephas


A collection of deep learning models for supervised detection of pull request clones on GitHub


A hierarchical deep learning model for supervised sentiment classification on diverse SE datasets


A suite of developer feedback and report generation tools to help improve the reviewability of GitHub pull requests


A neural network library using computational gate approach supporting common optimizers, activation and loss functions


An analysis framework to decode the social structure of developer mailing lists, IRC channels and Slack teams


Feature selection using semantic centroids calculated from the normalized correlation factors based on fuzzy c-means clustering

Get in touch!

Drop me an email or add me on LinkedIn.