Subhan Chaudry

Projects

Selected work

  1. 2026 · Cloud · Machine Learning · Featured

    Predicting Workload Instability in Cloud VMs

    Solo project · M.S. AI, UT Austin

    Forecasts whether a cloud VM will exceed 85% CPU within 15 minutes using workload telemetry. A Random Forest achieved 0.964 F1, with SHAP analysis identifying recent CPU peaks as the strongest predictors.

    • 99.66% Accuracy
    • 0.9999 ROC AUC
    • 0.964 F1
    Editorial cover: Predicting Workload Instability in Cloud VMs — 99.66% accuracy, 0.9999 ROC AUC, 0.964 F1.

  2. 2026 · Oracle Cloud · FinOps · Featured

    OCI Right-Sizer — Compute Rightsizing with Operator Control

    Cloud engineering · Tool development

    A Python CLI that turns OCI compute inventory and utilization telemetry into sizing recommendations and estimated cost differences. Operators review and approve changes, with live configuration checks before execution.

  3. 2025 · Medical Imaging · Deep Learning · Featured

    Improved Pneumothorax Classification using DCGAN Augmentation

    Solo project · M.S. AI, UT Austin

    Explores DCGAN-generated chest X-rays for balancing SIIM training data. A fine-tuned ResNet-18 reached 98.6% training accuracy; Grad-CAM and t-SNE visualizations help inspect model attention and learned representations.

    • 98.60% Train Accuracy
    • 1.000 ROC AUC
    Editorial cover: Improved Pneumothorax Classification using DCGAN Augmentation — 98.60% train accuracy, 1.000 ROC AUC.

  4. 2025 · NLP · Transformers

    Mitigating Answer-Length Bias in Fine-Tuned ELECTRA

    Solo project · M.S. AI, UT Austin

    Evaluates ELECTRA-small on SNLI and SQuAD, revealing lower exact-match scores for answers of six or more tokens. Oversampling improves long-answer F1 from 78.4% to 80.3%, with a trade-off in overall performance.

    • 90% SNLI Accuracy
    • 85.9% SQuAD F1
    • 78.1% SQuAD EM
    Editorial cover: Mitigating Answer-Length Bias in Fine-Tuned ELECTRA — 90% SNLI accuracy, 85.9% SQuAD F1, 78.1% SQuAD EM.

  5. 2023 · Accessibility · Computer Vision

    Eye Gazer — Webcam Accessibility Platform

    Accessibility · Machine Learning

    A browser-based communication board that uses webcam gaze tracking to let people with speech or mobility impairments select tiles and compose messages without specialized eye-tracking hardware.

    • Webcam-only Hardware
    Editorial cover: Eye Gazer — Webcam Accessibility Platform.

  6. 2020 · Consulting · Data Analysis

    Consulting Capstone with PEAK6 Investments

    MAP Consulting · B.B.A., Michigan Ross

    Built R analysis tools to evaluate growth and early-stage investments during a Michigan Ross MAP engagement with PEAK6. Delivered findings and due-diligence recommendations to firm executives.

    Editorial cover: Consulting Capstone with PEAK6 Investments.