Ishana Garuda
Computational Researcher & ML Engineer
Summary
- Computational researcher with 3+ years of ML and GPU acceleration focus with a B.S. in Computational Modeling and Data Analytics, seeking machine learning engineer or research scientist roles focused on efficient AI systems
- Technical expertise in LLM fine-tuning (LoRA, prompt-tuning), Parallel GPU programming (CUDA) and CPU programming (OpenMP), with strong communication skills for explaining complex topics
- Demonstrated ability to bridge technical depth—from optimizing CUDA kernels to creating interactive demos for non-technical users—while maintaining rigorous experimental standards
- Dean's List every semester, ASA DataFest Best Methodology, and consistent contributions to interdisciplinary research teams
Skills
Soft Skills
Experience
Research Assistant
Dr. Jason Wilson, Virginia Tech CMDA Department
- Developing data reduction methods using clustering (K-Means) and classification (kNN) for efficient large dataset analysis
- Re-engineering algorithms from OpenMP to CUDA, achieving 8.5x speedup (60s → 7s) on NVIDIA V100
- Conducting performance analysis with Nsight profiling tools to identify bottlenecks in kernel launch, memory transfer, and occupancy
- Creating reproducible benchmarks for the HPC community
Summer Research Intern
Dr. Chreston Miller, Virginia Tech Libraries & History Department
- Frameworked computational methods comparing historical legal documents' discussions of enslaved vs. white individuals
- Implemented LLMs via Vertex AI to modernize 17th century text while preserving historical context
- Applied embedding models with HDBSCAN, UMAP, and c-TF-IDF to identify themes across thousands of legal cases
- Created interactive visualizations (Plotly, Ipysigma) transforming raw data into actionable insights
Undergraduate Researcher
Dr. Chreston Miller, Virginia Tech Libraries
- Analyzed 15M tweets from Internet Archives using BERTopic and roBERTa sentiment analysis
- Validated findings via A/B permutation testing to ensure statistical significance
- Presented transformer-based models to interdisciplinary faculty as accessible insights
Undergraduate Learning Assistant — CMDA 3634
Virginia Tech CMDA Department
- Created C/OpenMP/CUDA API reference sheet for 40+ students
- Held office hours explaining thread synchronization, shared memory, race conditions, and memory coalescing
Academic Tutor & Peer Wellness Mentor
Virginia Tech APIDA+ Center
- Tutored calculus, linear algebra, statistics, data structures, and algorithms
- Created onboarding documents standardizing training processes
Projects
3D Heat Equation Simulation
C • CUDA • OpenGL • Nsight
Optimized CUDA kernels achieving 87% memory throughput and 98% GPU occupancy. Implemented CUDA-OpenGL interop for 60+ FPS volumetric rendering.
Parameter Efficient LLM Fine-Tuning
PyTorch • PEFT • HuggingFace
Prototyped Adaptive Sparse Fusion (ASF) framework dynamically fusing LoRA and Prompt-Tuning. Evaluated on GLUE benchmarks with mixed-precision training.
Named Entity Standardization
Python • Streamlit • OpenAI • RAG
Led development of LLM-based standardization system using RAG, deployed as Streamlit web app with semantic similarity search.
Quantum vs. Classical ML
Python • Qiskit • SVM
Modular package comparing Variational Quantum Classifiers and Classical SVMs with 100% code coverage.
Education
Virginia Tech
B.S. in Computational Modeling and Data Analytics
Minors: Computer Science, Mathematics, Statistics, Quantum Information Technology
GPA: 3.98 Major · 3.91 Cumulative
- Mathematical Modeling
- Data Analytics and Machine Learning
- Software Design
- Fourier Series & Partial Differential Equations
- Quantum Computing
- Applied Bayesian Statistics
- Experimental Design
- Linear Algebra
- Multivariable Calculus
- Probability & Statistics
- Data Structures & Algorithms
- Computer Organization
- Numerical Methods