Machine Learning & Data Science Lab, Kansas State University, Manhattan, KS
‣ Supervisor: Doina Caragea, Ph.D., Professor at Department of Computer Science, Kansas State University
‣ Crawled social media (Twitter, and Reddit) and analyzed huge amount of text to create NLP dataset for Stance Detection and establish baseline results using Supervised, Semi-supervised, and Zero-shot models
‣ Work on application of Machine/Deep Learning and Large Language Models (LLMs) on various domains including DNA Sequence Prediction, Social media Cyberbullying Perpetration Prediction, and Vehicle Crash Severity Prediction
Aug. 2019 – Dec. 2023
Graduate Teaching Assistant
Department of Computer Science, Kansas State University, Manhattan, KS
‣ Created Quizzes, graded labs/assignments, and academically supported students on the concepts of AI and Machine/Deep Learning for Introduction to Artificial Intelligence, Machine Learning & Pattern Recognition, and Deep Learning courses
‣ Graded homework and guided students through the concepts in Computer and Information Security, Database Systems, and Programming Languages courses
Aug. 2017 – Jul. 2019
Security Analytics Engineer
Logpoint Kathmandu, Nepal
‣ Researched, analyzed, and interpreted logs particularly from security and networking devices (such as proxies, firewalls, anti-virus products) for security monitoring, and as a data source for UEBA (User and Entity Behavior Analytics)
‣ Created, maintained, and optimized plugins, and scripts to correctly parse thousands of logs using Python’s Regex library
‣ Created analytics (Dashboards, Alerts, Reports), and correlation and detection rules, for incident management and monitoring threats and other activities via SIEM (Security Information & Event Management) solutions
‣ Developed Threat Intelligence, and Enrichment Plugins to fetch data from various feeds to correlate with parsed logs
‣ Took complete ownership of the development lifecycle for the product developement
Dec. 2016 – Feb. 2017
Software Development Internship
LeapFrog Technologies Kathmandu, Nepal
‣ Understand aspects of software development lifecycle with agile methodology by developing my own project (Video to Gif Converter), and website templates using HTML5, CSS, and JavaScript; maintain documentation as well when needed
‣ Communicate new ideas, issues, status, and outcomes in a collaborative environment of team and mentors
‣ Learn and apply version control tool (Git, Bitbucket), project management tool (JIRA), project level technologies and code review process
Publication / Presentation
Jan. 2024
(Paper Presentation): Gyawali, Nikesh, Sarthak Khanal, Doina Caragea, H M Abdul Aziz, and Eric J. Fitzsimmons. “Predicting Commercial Motor Vehicle Crash Severity in Kansas at District Level using Explainable Machine Learning.” 103rd Transportation Research Board Annual Meeting. 2024.
Sep. 2023
(Publication): Gyawali, Nikesh, et al. “Using recurrent neural networks to detect supernumerary chromosomes in fungal strains causing blast diseases.” bioRxiv (2023): 2023-09. (paper link)
Sep. 2020
(Publication): Tasali, Qais, Nikesh Gyawali, and Eugene Y. Vasserman. “Time series anomaly detection in medical break-the-glass.” Proceedings of the 7th Symposium on Hot Topics in the Science of Security. 2020. (paper link)
Technical Skills
Programming Languages: Python, Bash, JavaScript Databases: MySQL, MongoDB (NoSQL) Frameworks and libraries: PyTorch, Keras, TensorFlow, Scikit-learn Tools & Technologies: CUDA, Regex, Amazon Web Services (AWS), Django, HTML5, CSS, Bootstrap Others: Object-Oriented Programming, Slurm, High-Performance Computing (HPC), Large Language Models (LLMs)
Selected Projects
1. Commercial Motor Vehicle crash severity prediction(Aug. 2022 – Ongoing)
Project in collaboration with Civil Engineering Department at Kansas State University and Kansas State Department of Transportation to develop ML models to predict crash severity in Commercial Motor Vehicles (CMVs)
Utilized various Machine Learning models (e.g., Random Forest, Gradient Boost, XGBoost, CatBoost) including Transformer based models (TabNet) on tabular dataset for severity classification
Created parser to parse crash pdf forms using Deep Learning OCR models and created analytics (dashboards and reports) to visualize the data
2. FunBERT: A pre-trained BERT model for DNA language in Fungal Genome (Aug. 2023 – Ongoing)
Project to pre-train a Transformer model from scratch on Fungal Genomic sequence for various downstream tasks such as Sequence classification, Missense variant effect prediction, and Important sequence motifs identification
3. Public Perception of Vaccines: Before and After the COVID-19 Outbreak (Dec. 2022 – May 2023)
Project to understand how people talk about Vaccines in social media
Crawled 10 years (2013-2022) of tweets data related to vaccines and used Lexicon based analysis to study the polarity and perception shift of people over the years
