Parallel Triangle Counting
Designing scalable algorithms for counting triangles and related subgraph patterns in large-scale graphs.
Ph.D. Student in Data Science · NJIT
Atieh Barati-Nia
Graph Algorithms · High-Performance Computing · Dynamic Graph Analytics
I design and evaluate efficient graph algorithms, with current work on parallel triangle counting in large-scale graphs, dynamic graph analytics, and performance-aware scientific software.
Current focus Parallel and dynamic triangle counting at scale
Recognition IEEE HPEC 2025 Outstanding Student Paper
Research areas
Graphs · Systems · AI-assisted software
About
Research grounded in practical performance.
I am a Ph.D. student in Data Science at the New Jersey Institute of Technology, advised by Prof. David A. Bader. My research focuses on scalable graph algorithms, parallel and batch-dynamic triangle counting, dynamic graph analytics, and performance engineering for scientific software.
I study the implementation details that shape real performance, including data layout, memory traffic, cache behavior, parallel scalability, and rigorous experimental evaluation. I also investigate how modern language models generate and optimize low-level graph code.
Beyond research
I love board games, solving mysteries, and traveling. Just as I enjoy technical challenges and learning new ideas, I also like trying unfamiliar activities, exploring new places, and taking on something new for fun.
Research
Current directions
Algorithm design, parallel implementation, and empirical evaluation for large graph workloads.
AI for Efficient Software
Evaluating whether language models can generate correct and efficient systems code.
Selected projects
Code and experiments
Representative work from graph analytics, machine learning, and empirical systems research.
LLM-Generated Code for Graph Analysis
Evaluating generated C code across compilability, correctness, runtime, memory footprint, and integration readiness.
RAG + Temporal Graph Networks
Combining retrieval-augmented generation with temporal graph learning for evolving graph data.
Fraud Detection with Deep Models
Comparing permutation-invariant and transformer-based tabular models with XGBoost.
Hate-Speech Text Representations
Comparing TF-IDF, Word2Vec, and fine-tuned BERT for hate-speech detection.
Featured publication
Recent work
IEEE HPEC 2025
Evaluating Efficiency and Novelty of LLM-Generated Code for Graph Analysis
A systematic evaluation of language models generating efficient C implementations for graph-analysis routines.
Outstanding Student Paper Award
Contact
Research and collaboration
For questions about my work or potential collaboration, please get in touch.