resume.md [OK]

Saadhya Vijayvargiya

Computer Science Honors Student / AI Systems Builder / Software Engineer

Computer Science Honors student at The University of Texas at Austin with experience in AI model development, full-stack platforms, applied ML research, and competitive programming.

education 01

$ cat education.log

The University of Texas at Austin / B.S. Computer Science Honors
  • Relevant coursework: Vector Calculus (H), Data Structures (H), Discrete Math (H), Business Communication (H), Computer Architecture (H).
  • (H) indicates honors coursework.
experience 04

$ tail -f experience.log

AI Development Intern / Alpha AI
  • Designed and fine-tuned a 7B-parameter language model for competitive programming.
  • Constructed a pipeline to process web-scraped data into labeled datasets for model development.
  • Fine-tuned with QLoRA, reducing training memory by 65% compared with full fine-tuning.
Software Engineering Intern / Students Teaching Finance
  • Architected and deployed a full-stack Node.js platform, enabling onboarding that fueled a 4x increase in international chapter launches within 12 months.
  • Engineered scalable infrastructure supporting 100+ global chapters and approximately 4,000 active student users.
Research Assistant / UC Berkeley
  • Researched the intersection of AI and legal sentencing, developing a machine learning model achieving 83% accuracy on in-distribution cases and 71% accuracy on out-of-distribution datasets.
  • Engineered model architecture that reduced bias in sentencing predictions by 34% compared to the baseline approach.
  • Processed and analyzed over 2M anonymized federal and state sentencing records to train and validate models.
Research Fellow / Nontrivial Fellowship
  • Investigated quantum-inspired neural architectures to reduce training energy consumption, achieving a 28% lower compute cost on MNIST and CIFAR-10 benchmarks.
  • Implemented a quantum kernel classifier using Qiskit's Aer simulator, outperforming classical SVMs by 14%.
projects 02

$ ls ./projects

Hastorium / C++, NumPy
  • Built an AI-powered telemetry analysis system that processes high-performance computing output to develop real-time strategies for F1 racing.
  • Engineered modular C++ architecture with a regression model for generative analysis and ElevenLabs text-to-speech, enabling actionable performance recommendations 43% faster than the previous baseline.
TraceTune / Python, Git, PEFT, Hugging Face
  • Developed a pipeline that converts LLM conversation transcripts into supervised fine-tuning datasets by extracting user preferences, formatting conventions, and behavioral patterns.
  • Distilled the dataset into a small model that learns conventions and styles more effectively; fine-tuned with LoRA to reduce token usage per session by roughly 28%.
achievements 03

$ grep awards achievements.txt

  • American Computer Science League - 1st Place.
  • USACO Platinum Division Contestant.
  • International Collegiate Programming Contest SCUSA - 7th place.
additional information 05

$ languages

Java Python C/C++ SQL JavaScript TypeScript HTML/CSS LaTeX Swift Go Perl R

$ frameworks

Next.js Tailwind TensorFlow Pandas PyTorch Oracle Greenfield Radius Azure CUDA MATLAB

$ interests

Algorithm Design Machine Learning Applied Mathematics Docker and Kubernetes Competitive Programming Data Science Statistics Computer Vision Deep Learning Quantitative Finance