Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

M.S. Computer Science graduate student at Wayne State University focusing on machine unlearning and parameter-efficient fine-tuning for LLMs. Researches constrained entropic unlearning and builds end-to-end pipelines to selectively forget target data while preserving performance on retain sets. Previously built real-world full-stack applications and RPA automations, and supported students as a teaching assistant and tutor in software and systems courses. Strong hands-on experience with PyTorch/TensorFlow, Transformers, CUDA, and modern web development using React/Next.js and API-based backends.…M.S. Computer Science graduate student at Wayne State University focusing on machine unlearning and parameter-efficient fine-tuning for LLMs. Researches constrained entropic unlearning and builds end-to-end pipelines to selectively forget target data while preserving performance on retain sets. Previously built real-world full-stack applications and RPA automations, and supported students as a teaching assistant and tutor in software and systems courses. Strong hands-on experience with PyTorch/TensorFlow, Transformers, CUDA, and modern web development using React/Next.js and API-based backends.

Vijayalakshmi Sivaraman

Software Engineer, Programmer, Data Scientist, +2





M.S. Computer Science graduate student at Wayne State University focusing on machine unlearning and parameter-efficient fine-tuning for LLMs. Researches constrained entropic unlearning and builds end-to-end pipelines to selectively forget target data while preserving performance on retain sets. Previously built real-world full-stack applications and RPA automations, and supported students as a teaching assistant and tutor in software and systems courses. Strong hands-on experience with PyTorch/TensorFlow, Transformers, CUDA, and modern web development using React/Next.js and API-based backends.…M.S. Computer Science graduate student at Wayne State University focusing on machine unlearning and parameter-efficient fine-tuning for LLMs. Researches constrained entropic unlearning and builds end-to-end pipelines to selectively forget target data while preserving performance on retain sets. Previously built real-world full-stack applications and RPA automations, and supported students as a teaching assistant and tutor in software and systems courses. Strong hands-on experience with PyTorch/TensorFlow, Transformers, CUDA, and modern web development using React/Next.js and API-based backends.

Available to hire

M.S. Computer Science graduate student at Wayne State University focusing on machine unlearning and parameter-efficient fine-tuning for LLMs. Researches constrained entropic unlearning and builds end-to-end pipelines to selectively forget target data while preserving performance on retain sets.

Previously built real-world full-stack applications and RPA automations, and supported students as a teaching assistant and tutor in software and systems courses. Strong hands-on experience with PyTorch/TensorFlow, Transformers, CUDA, and modern web development using React/Next.js and API-based backends.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Intermediate

Intermediate

Intermediate

Intermediate

Language

Javanese

Advanced

Work Experience

Experience Library Assistant at Wayne State University

January 1, 2026 - Present

Handled circulation desk and patron support, collection organization, and opening/closing operations across library functions.

Real-Time Multiplayer Quiz Platform (Project)

January 1, 2026 - Present

Built a real-time paired multiplayer quiz system with Next.js, TypeScript, Tailwind CSS, and Framer Motion, backed by Supabase PostgreSQL Realtime subscriptions and deployed to Vercel. Implemented paired mode where both players answer and predict each other’s responses; Supabase Realtime detects joint submission and routes players to a results page with mutual scoring.

Research (Machine Unlearning) / Graduate Research at Wayne State University

January 1, 2026 - Present

Reproduced Constrained Entropic Unlearning (NeurIPS 2025) by building a Python pipeline over a decoder-only transformer trained on TOFU. Replaced fixed scalarization with constrained optimization and implemented logit-margin flattening loss; compared Gradient Ascent, Gradient Difference, and PDU. Conducted controlled experiments to establish memorization thresholds and minimum scale requirements for separable unlearning, extending to GPT-2 small (124M) fine-tuned on TOFU to demonstrate selective forgetting while preserving retain-set fluency.

Teaching Assistant at Irvin D. Reid Honors College, Wayne State University

January 1, 2025 - Present

Supported students in game development and JavaScript engine implementation; provided technical guidance and debugged projects across the semester.

Home Cleaning Service Management System (Project)

January 1, 2025 - Present

Built a full-stack web application with React.js, PHP, and SQL featuring role-based authentication, session encryption, and audit logging. Designed a relational schema and implemented CRUD operations for booking lifecycle, status tracking, and secure data retrieval with input validation and access control.

Research (Invariant Risk Minimization / Distribution Shift Study)

January 1, 2025 - Present

Compared ERM, IRM, and IRM-v1 using Colored-MNIST with MLP feature extractors in PyTorch with CUDA across three controlled distribution-shift environments. Found IRM improved out-of-distribution accuracy by 31% over ERM, which collapsed by 70% under inverted correlations, supporting invariant feature learning.

Research (Parameter-Efficient Fine-Tuning) / LoRA Implementation

January 1, 2025 - Present

Implemented LoRA fine-tuning on DistilBERT for IMDB sentiment classification using Hugging Face Transformers and PyTorch. Ran comparative trials against GPT-2 and BERT, reducing trainable parameters by over two orders of magnitude while cutting training time by ~60% and GPU memory by ~40%, maintaining full fine-tuning accuracy across scaled runs.

Tutor at Wayne State University

January 1, 2024 - January 1, 2025

Tutored undergraduate students in Operating Systems, Computer Networks, Computer Architecture, Web Technology, Python, and Probability & Statistics.

Student Developer Champion (RPA Developer) at UiPath

January 1, 2022 - January 1, 2023

Built and deployed RPA automation bots using REFramework and Python with API integrations across Slack, Discord, Chrome, Excel, and Word. Deployed automations on AWS and Heroku and mentored peers on automation lifecycle.