Machine Learning Engineer Intern

Gaussian-Centered Cropping for Contrastive Learning – Computer Vision, PyTorch, Unsupervised Learning, CNNs

• Proposed Gaussian-Centered Cropping (GCC) to reduce false positives in SSL.
• Achieved +2.7–12.4pp accuracy vs. RandomCrop with no extra compute.
• Preprint: Taming the Randomness – [arXiv link]

Multimodal Sentiment Analysis via Emotional Embeddings – NLP, LLM, Transformers, LSTM

• Leveraged emotional embeddings from GoEmotions to classify Yelp reviews.

• Demonstrated transfer learning preserved sentiment signals across domains.

• Power Systems: Testing & Commissioning, HV/MV/LV Systems, CT/VT

• Protection Systems: Relay Configuration (ABB, GE, Schneider, Alstom), SCADA

• Data & Modeling: SQL, MySQL, Feature Engineering, Statistics, Predictive modeling, Python (NumPy, Pandas, SciPy), R

• RAG & LLMs: LangChain, ChromaDB, LlamaIndex, Hugging Face Transformers

• Model Deployment & MLOps: FastAPI, Docker, Kubernetes, AWS Cloud (SageMaker, EC2, S3), MLflow, Git, CI/CD pipelines

• Multi-modal AI: Computer vision, NLP, audio processing, 3D meshes/point clouds

• Visualization: Tableau, Plotly, Matplotlib, Seaborn