Akhil Nampally
B.S. Mathematics + Computer Science · Georgia Institute of Technology
Hi! I'm an undergrad at Georgia Tech studying CS and Math. I'm currently a researcher in the ViTAL Lab under Dr. Hyeokhyen Kwon, working on behavior detection in adults with high-functioning autism. I'm also an incoming Data Engineer Intern at Meta this summer. My interests span AI/ML, deep learning, and software development.
Research
Current
ViTAL Lab · Dr. Hyeokhyen Kwon
Working on the EAC project, which aims to detect and understand challenging behaviors in adults with high-functioning autism. My role involves exploring how sensor data and machine learning can be used to build tools that support better care and intervention for this population.
Past
Amphibian Specimen Segmentation via DINOv2
Used DINOv2's self-supervised vision transformers to identify and segment amphibian specimens from specialized biological image datasets. The project explored how self-supervised learning can work well in niche classification tasks where labeled data is scarce. This work is currently being written up as a paper.
Living Dynamical Systems Lab · Prof. Simon Sponberg
Worked on building a real-time data acquisition system for a capacitive liquid level sensor, then used Python to develop predictive models and visualizations to make sense of the sensor data. A big part of the work was automating the data logging and calibration process, which ended up cutting calibration time significantly.
Interests
- Assistive technology & accessibility — using ML to improve quality of life for people with disabilities
- Self-supervised & efficient learning — reducing labeled data requirements and compute overhead
- Large language models — quantization, fine-tuning, and deployment at scale
Projects
LLM-SDQ — Efficient Large Language Models
A project around making large language models more efficient. I implemented switchable-precision quantization on GPT-2 so different layers could run at different bit-widths, which cut GPU memory usage by 4×. I also added LoRA adapters and fine-tuned on SQuAD, seeing a solid bump in F1 score over the baseline.
PhantomShield — Real-Time Deepfake Detection
A real-time deepfake detection tool built for live video calls. I trained a ResNet18-based model to detect deepfake video with 85% accuracy, then wrapped it in a cross-platform desktop app using Electron.js that shows live alerts and a risk indicator during calls.
SlangBERTa — Slang Emotion Classification
Fine-tuned RoBERTa to recognize emotions in slang-heavy, informal text — the kind of language standard NLP models tend to struggle with. I built a synthetic data pipeline to cover rare emotions and edge cases, then spent time analyzing where the model was still getting things wrong.
EV-Arb-Tool — Odds Evaluation & Prop Finder
A full-stack tool for finding positive expected value bets across sportsbooks and DFS platforms. It pulls in odds, calculates fair probabilities, and surfaces high-value opportunities automatically — processing over 10,000 props a day at its peak.
Resume
Download Resume (PDF)Contact
Best way to reach me: anampally3@gatech.edu