I am an Engineer specialized in AI.
Previously, I was invited to attend a special in-person Educational Program for the “Gifted Youth” at Stanford University, California in Cosmology.
I was awarded two bachelor’s degrees simultaneously - B.S. in Electrical and Computer Engineering and B.S. in Atmospheric Sciences both at the University of Illinois at Urbana-Champaign where I was advised by Erhan Kudeki and Eric Snodgrass.
Subsequently I worked as an engineer at Google in Mountain View, California where I built models and systems to solve problems at the intersection of Science, technology and data (GoogleX).
I am currently pursuing a master’s in AI with a focus on NLP + RLHF in the UK.
This project explores the extent to which linguistic and cultural biases present in training corpora and preference dataset shape LLM responses during inference. By examining the influence of individual pre and post-training datasets, we investigate whether models reflect or reinforce cultural norms without deliberate intent.
Designed and trained novel convolutional neural networks (CNNs) using PyTorch for object detection and defect classification in cables. Focused on model accuracy, feature maps, and error detection. Applied deep learning to automate visual inspection and improve quality control.
A robust subscription management system for GCP services built with gRPCs and protobufs. This project included liasing with external vendors to fulfill custom Google cloud infrastructure requirements for enterprise clients - providing scalable billing solutions for a seamless customer experience.
The work on the Pixel Phones team involved design and implementation of a novel 'eVNA' API in collaboration with industry partners such as STMicroelectronics and National Instruments. Analysis of data informed optimizations in the consumer hardware manufacturing pipelines to enhance product development. The API has been included in all Pixel phones produced since its inception enhancing connectivity and the consumer hardware experience for millions of Pixel phone users worldwide to date.
The Pixel Buds project required implementation of complex calibration algorithms for user-critical technologies such as Bluetooth LE and IoT (Zigbee). The codebase was designed to ensure performance and compliance with requirements that vary by country and region. Analyzing large-scale datasets proved crucial for improving product quality throughout the product development. I worked on the core team throughout the product lifecycle from initial design to launch.
This research presented at the American Geophysical Union utilized advanced resonance lidar systems at the Arecibo Observatory to analyze mesospheric sodium (Na) variability and its correlations with geomagnetic activity and solar flux. The data and models provided insights into seasonal and aperiodic variations in atmospheric composition.
This project used radar data observed at the Arecibo Observatory to model how powerful radio signals (Incoherent Scatter Radar) can change the ionosphere, the research revealed new insights into the ionosphere and its behavior.