- Developed and trained small LLMs to run on on-device in resource-constrained systems to aide in datacenter networks in anomaly management and remediation- Orchestrated DDP Training with Deepspeed over HPU clusters for fine-tuning in PyTorch- Engineered a bespoke RAG system for Cisco networking products- Predictive modeling of network topology- Improving observability into Data-centers with React, d3.js

• Developed the UI for an internal Cisco hardware resource management web application using React• Application allowed developers to lease required hardware for testing and could track usage between business units• The project freed developers’ workflow, sped up the time it took to acquire hardware to test, identified, flagged, removed offline systems, and gave insights into usage and hardware utilization between business units

• Collaborated on Python Software that quantified localized Calcium Spikes of endothelial cells• Classified extracted image features correlated with the phenotype present in heart cancer using PCA with k-means• Proposed new methods for cleaning image data to avoid misclassification of fluorescent intensities• Utilized data visualization techniques to create figures that are cited in upcoming publication

• Performed dimensional reduction on reticulin stained image-features with T-SNE unsupervised learning algorithms to cluster and find insights in the data• Implemented StarDist deep learning method in Python to do dense nuclear segmentation of tissue microarrays• Constructed an image fingerprint for the CNN pipeline to identify biomarkers and predict ovarian cancer diagnosis• Wrote a description of data cleaning and preprocessing techniques used that will be in cited in upcoming… Show more • Performed dimensional reduction on reticulin stained image-features with T-SNE unsupervised learning algorithms to cluster and find insights in the data• Implemented StarDist deep learning method in Python to do dense nuclear segmentation of tissue microarrays• Constructed an image fingerprint for the CNN pipeline to identify biomarkers and predict ovarian cancer diagnosis• Wrote a description of data cleaning and preprocessing techniques used that will be in cited in upcoming paper- Implemented Python scripts to anonymize patient slide information and de-artifact tissue slide images Show less

• Led the social committee and coordinated with the VP of Finance to better budget and plan for events to improve fraternity sorority relations• Conducted screenings and interviews of over 1,000 applicants for our Balanced Bruin Annual Scholarship program.

• Trained T-SNE clustering and ARIMA predictive models to forecast near-misses, traffic volume, and incidents in Phoenix street corridors that had LIDAR/camera sensors• Deployed models to Azure ML Studio to be used in a real-time web application to prevent accidents and report data to city officials