Machine Learning Engineer building efficient computer-vision systems for object detection, segmentation, and real-time deployment.
MS in Computer Science at Oregon State University and ML Engineer at Microtec Inc. I work on object detection and segmentation, automated data-labeling pipelines, and optimizing models for edge deployment.
- 🎓 MS in Computer Science, Oregon State University (2023–2025)
- 💼 Machine Learning Engineer, Microtec Inc
- Built an assisted annotation tool (PySide6 + SAM) that doubled labeling throughput and replaced legacy tooling in production.
- Architected an automated labeling system (object detectors + SAM), cutting manual labeling effort by ~50%.
- Trained a foundational detector on 300k+ images / 20 classes to 87% mAP.
- Designed novel loss functions and pretraining methods, reducing false positives and data requirements by 30%.
- Thesis: Evaluating Thumbnail Preserving Encryption(TPE)'s vulnerability to Transformers. Identified security parameter required to make TPE resilient to SoTA super-resolution attacks.
- Conv-MAE: MAE-style masked pretraining for a hybrid conv-transformer (QT-ViT), reaching 80.18% on ImageNet-100.
- Conv-MAE — self-supervised MAE pretraining for hybrid vision backbones.
- Brain-Tumor-Segmentation — comparative 2D vs 3D UNet study on the BraTS dataset.
- Network-Anomaly-Detection — autoencoder-based anomaly detection on NSL-KDD.
- Languages: Python, C++, SQL
- ML / DL: PyTorch, CUDA, ONNX, TensorRT, OpenMMLab, FFCV
- Tooling: Git, Docker, Azure DevOps, Jenkins, MLflow, Valohai
- Other: PySide6, OpenCV, NumPy, Pytest