Hi! I'm Aditya, and I like playing with computers and creating intelligent machines! As an undergrad at Georgia Tech, I major in Computer Science.
Over the years, I've worked on several projects doing research and building software. Much of this has been work in AI, spanning domains from healthcare to robotics.
Outside of computing, I'm very interested in spaceflight, entropy maximization (extremely high-effort ethical pranksterism), writing bad science fiction, and much more.
EducationOver the years, I've worked on several projects doing research and building software. Much of this has been work in AI, spanning domains from healthcare to robotics.
Outside of computing, I'm very interested in spaceflight, entropy maximization (extremely high-effort ethical pranksterism), writing bad science fiction, and much more.
Georgia Institute of Technology
Bachelor of Science (BS) in Computer Science
May 2025
Final year CS major, with concentrations in AI and theory.
GPA: 3.94/4.00, Faculty Honors & Dean's List
Carnegie Mellon University: Machine Learning Intern
Sep. 2022 - Present
- Introduced and demonstrated a novel GAN-based deep learning approach for the first-ever synthetic construction of cardiac motion from electrical signals (ECG).
- Implications include a 25x cost reduction for 59% of cardiac ultrasounds, 19% more reliable diagnosis vs. current approaches, and unlocking clinician interpretability for complex AI model predictions.
- Recognized by the American Society of Echocardiography as one of America's top four researchers under the age of 40.
- Submitted a first author paper - stay tuned!
Georgia Institute of Technology: Student Researcher
April 2024 - Present
- Under Dr. Animesh Garg at PAIR lab, developing a geometrically-aware approach to LLM-based reward function design and skill learning for applications in robotics and robotics foundation models.
- Created a language-grounded vision pipeline for real-time object manipulation in both simulation and real world.
Hewlett Packard Enterprise: Intern
May 2023 - Aug. 2023, Jun. 2024 - Aug. 2024
- Architected and built a series of applications and deep learning models for processing legal text.
- Developed an unsupervised, LLM-based pipeline to train specialized, efficient models across diverse legal problems.
- Enabled combined savings of $1M+ annually, previously allocated for a specialized IP legal team.
- At 98%+ accuracy, marked the company's first success; now a trade secret used worldwide as production software.
PPC Pharmaceuticals: Intern
June 2022 - Aug. 2022
- Led the statistical analysis for a clinical trial of a pharmaceutical product aimed at mitigating antibiotic misuse.
- Developed new disease-agnostic severity metrics to compare and test the effectiveness of different drug variants and placebos across various illnesses.
- First author in publication to Infection and Drug Resistance.
Carnegie Mellon University: Machine Learning Intern
July 2020 - Feb. 2021
- Developed machine learning-based algorithms to introduce the novel capability of easily and reliably discriminating between COVID-19 and influenza, a key frontline diagnostic need for hospitals during the pandemic.
- Using only vital signs, achieved AUC scores of 97%+ on external test sets, demonstrating exceptional reliability.
- Co-authored a publication to npj Digital Medicine.
Independent Student Research
Zoolbot & SCAMADER: Deep Learning for Wildlife Conservation
June 2018 - July 2020
- Developed a new deep meta-learning approach for one-shot learning in computer vision, applied to the identification of extremely rare species as an alternative to expensive, slow, resource-intensive DNA-based testing.
- Achieved 80% accuracy in species classification using CNNs developed from merely one image per species.
- Several recognitions, including Grand Award at Intel’s ISEF, USAID’s Digital for Development award, and Google Science Fair’s Asia-Pacific regional award.
Independent Student Research
FlareNet: Applying Deep Learning in Heliophysics
July 2019 - May 2020
- Mentored by Dr. Dibyendu Nandi, developed a new predictive model for space weather, enabling neural nets to make long-range predictions, having vital applications in protecting critical systems.
- ConvLSTM-based encoder-decoder models of solar magnetic fields yielded at least 80% accuracy in CME and solar flare prediction.
- Finalist at Regeneron’s International Science Fair and Grand Award at Broadcom’s IRIS.
The worm on your screen is using the "brain" of Caenorhabditis elegans, a roundworm. If I'm being overly dramatic, this webpage is sort of like the Matrix, but for worms.
More specifically, your computer is currently simulating the C. elegans connectome in real time, using simple perfect intergrate-and-fire-type neurons. After a random initialization, the worm is frequently given arbitrary chemosensory stimuli (sensing virtual food) at random intervals as well as tactile feedback when it hits the walls of the webpage. Its muscular reactions (in the form of squiggling motions) provide locomotion, here mapped through a simple singular velocity computation.
The code and connectome are based on the open-source OpenWorm project, an attempt at "building the first digital life form". (More specifically, I use the C. elegans robot project.)
Download ResumeMore specifically, your computer is currently simulating the C. elegans connectome in real time, using simple perfect intergrate-and-fire-type neurons. After a random initialization, the worm is frequently given arbitrary chemosensory stimuli (sensing virtual food) at random intervals as well as tactile feedback when it hits the walls of the webpage. Its muscular reactions (in the form of squiggling motions) provide locomotion, here mapped through a simple singular velocity computation.
The code and connectome are based on the open-source OpenWorm project, an attempt at "building the first digital life form". (More specifically, I use the C. elegans robot project.)
Dark Theme
Contact
Feel free to send me an email at adityark@gatech.edu!