TypeSpaceBERT: A Deep Similarity Learning-based CodeBERT Model for Type Inference
Fine-tuning a large natural language model for the task of type inference in TypeScript, centered on similarity heuristics in Euclidean spaces. Paper.
Projects
A Survey of Test Case Generation Strategies for RESTful APIs
Surveying, contrasting, and analyzing the current body of research in the field of automated test case generation for REST APIs. Paper.
In pursuit of reproducible release engineering pipelines for extensible machine learning based systems
Transforming a Jupyter Notebook ML application into a production-ready microservice-based system with extreme reproducibility and developer friendliness in mind. Paper. Code.
The effects of Local Search on Random Key encoding schemes for the Traveling Salesperson Problem
Implementing a novel evolutionary local search heuristic for the TSP problem and comparing its effects on three permutation-preserving encodings against the standard 2-opt heuristic. Paper.
Using graph addition-contraction to compute tighter bounds in hybrid MaxSAT/CP solving for Correlation Clustering
Using a MaxSAT-based approach for solving the correlation clustering problem exactly, and comparing state of the art implementations from literature. Paper.
Modeling System Behaviour from Log Analysis Using Meta-Heuristic Search
Proposing, implementing, and evaluating four novel evolutionary to model the behaviour of software systems based on their logs. Paper. Code.