cv
Contact Information
| Name | David N. Palacio |
| Professional Title | Ph.D(c) Computer Science |
| danaderp@gmail.com | |
| Phone | (757) 279-4265 |
| Location | , Williamsburg, Virginia VA 23185 |
| Website | https://danaderp.github.io/danaderp/ |
Professional Summary
Ph.D. candidate specializing in deep learning and software engineering, with expertise in applying causal inference techniques to interpret large language models (LLMs), aimed at automating software maintenance tasks. Eager to leverage these skills as a Research Scientist focused on AI development, trustworthiness, alignment, and explainability.
Experience
-
2021 - 2022 Research Intern
Microsoft
Published a patent for an innovative debugging tool designed to provide recourse to practitioners in explaining deep generative models trained on code as data. Worked in collaboration with four Microsoft Senior research scientists.
- Post Hoc Interpretability
-
2020 - 2020 PhD Intern
Cisco Systems
Investigated an information theory approach for interpreting and evaluating software retrieval techniques.
- Information Theory, Traceability
-
2016 - 2016 Software Engineer (Senior Back‑end)
KSMTI
Engineered reactive and functional programming architectures for enabling fast development of any type of marketplace business.
- Scala, Functional Programming
Education
-
2017 - 2024 Williamsburg, VA
Awards
-
2014 Indian Government Scholarship in Noida, India
Indian Government
Publications
-
2024 Toward a Theory of Causation for Interpreting Neural Code Models
IEEE Transactions of Software Engineering
It concerned a causal theory for understanding LLMs for code.
-
2022 Debugging Tool for Code Generation Neural Language Models
United States Patent
A debugging tool identifies the smallest subset of an input sequence or rationales that influenced a neural language model.
-
2024 Which Syntactic Capabilities Are Statistically Learned by Masked Language Models for Code?
ICSME
Limitations of evaluating Masked Language Models (MLMs) in code completion tasks.
Skills
Deep Learning for Software Engineering (PhD): PyMC3, PyTorch, Foundation Models, LLMs for Code, Unsupervised Models, Code Generation
Languages
Spanish : Native speaker
English : Fluent
German : Intermediate
Interests
LLMs for Code: Explainability, Interpretability, Trustworthiness, Causal Inference, Representation Learning, Data Analysis
Certificates
- Distributed Machine Learning with Apache Spark - Berkeley University (2016)