Neuro-AI Researcher

KushalKapoor

Building the future of neuro-inspired AI

PhD-bound researcher developing hippocampal-inspired architectures for continual learning. Bridging computational neuroscience and deep learning to create more intelligent systems.

01. About Me

Researcher at the intersection of neuroscience and AI

I'm a Computer Science student at the University of Maryland, College Park, pursuing a B.Sc. in Computer Science Honours (Machine Learning) and Neurotechnology (Individual Studies), with a minor in Computational Finance.

My research centers on developing hippocampal-inspired architectures for continual learning. I'm particularly interested in how biological principles like pattern separation in the Dentate Gyrus can address catastrophic forgetting in neural networks.

Beyond research, I've worked as a Software Research Engineer at Marriott International and lead the quantitative desk at Apex Fund, where I develop AI-powered trading strategies.

I'm actively seeking PhD opportunities for Fall 2026 to continue advancing the field of neuro-inspired artificial intelligence.

Technical Skills

PythonC++MATLABPyTorchJAXTransformersDiffusion ModelsContinual LearningEvolutionary AlgorithmsMEG/EEG AnalysisSpike Encoding ModelsHippocampal ArchitecturesDockerLinuxGit

Neuro-AI Research

Developing hippocampal-inspired architectures for continual learning systems

Computational Neuroscience

Bridging biological intelligence with artificial neural networks

Deep Learning

Expertise in PyTorch, TensorFlow, and large-scale model development

PhD Bound

Seeking Fall 2026 opportunities in Neuro-AI and Computational Neuroscience

Relevant Coursework

Molecular and Cellular NeuroscienceSignal ProcessingMolecular NeuroethologyMachine LearningNeural Systems and CircuitsFMRI AnalysisBioPsychologyApplied Statistics

01.5 Research Experience

Advancing computational neuroscience and AI

Perception and Robotics Group

Undergraduate Research Assistant

March 2025 – Present
•
College Park, MD
  • ▸Led development of HiCL, a hippocampal-inspired continual learning architecture using DG-gated Mixture-of-Experts with sparse pattern separation (top-k sparsity at 5%) and cosine similarity routing to prototypes, achieving over 95% routing accuracy without explicit gating networks under Prof. Yiannis Aloimonos.
  • ▸Designed a dual-phase training strategy integrating Elastic Weight Consolidation, prioritized replay, and contrastive losses for memory consolidation, yielding robust performance on Split CIFAR-10 (92.38% Task-IL accuracy) and Split Tiny-ImageNet (62.76% Task-IL) at 5% of baseline compute costs (e.g., 38.21 MFLOPs for small model).
  • ▸Conducting an Honors thesis on high-dimensional vectors and embeddings. Attended advanced seminars and guest lectures in neuro-inspired AI to deepen interdisciplinary expertise.
  • ▸Finetuned Atlantis to recreate underwater scenes using components from SONAR emissions and text descriptions.
Continual LearningEvolutionary AlgorithmsComputer Vision

Shamma Lab

Undergraduate Research Assistant

March 2024 – November 2024
•
College Park, MD
  • ▸Developed a computational pipeline to decode speech from MEG neural recordings under Dr. Shihab Shamma
  • ▸Implemented spike-based encoding models to map temporal neural patterns to individuals reading poems.
  • ▸Fine-tuned GPT to predict neural response patterns from speech stimuli
MEGNeural DecodingNLP

Genome Computing Lab in FIRE

Undergraduate Research Assistant

August 2022 – December 2023
•
College Park, MD
  • ▸Developed 3D models of nucleosome-decorated DNA fragments to investigate linker sequence effects on structural conformation and rigidity, utilizing 3DNA, Python and PyMOL for optimization and visualization of datasets from the Todolli database.
  • ▸Analyzed sequence-dependent DNA-protein interactions in lac repressor loops, to reveal multistate variance patterns and binding site influences.
  • ▸Managed large-scale biological data processing, implementing scripts for data retrieval, manipulation, and clustering
  • ▸Presented research poster 'Linker DNA Sequence Effects on Nucleosome Orientation' at FIRE Research Day
Computational BiologyStructural BiologyData Science

Presentations

Upcoming

HiCL: Hippocampal Inspired Continual Learning

The Fortieth AAAI Conference on Artificial Intelligence (AAAI-26) • January 2026

Poster Presentation

Invited Talk

HiCL: Hippocampal Inspired Continual Learning

DEVCOM Army Research Laboratory, University of Tennessee, Knoxville • July 2025

Presented to Professors, Senior faculty, graduate students, lab members and SURF students

Poster

Linker DNA Sequence Effects on Nucleosome Orientation

FIRE Research Day, University of Maryland • November 2023

04. Publications

Research contributions

Published Work

HiCL: Hippocampal-Inspired Continual Learning

AAAI 2026 (Accepted)In Print

Kapoor, K., Mackey, W., Aloimonos, Y., & Lin, X. (2025). arXiv preprint arXiv:2508.16651.

Continual LearningNeuro-AIDeep Learning

Preprints & Work in Progress

Pattern Separation for Nonlinear, Locality-Preserving Factor Orthogonalization

NeuroAI Workshop @ AAAI 2026 (Under Review)

Kapoor, K. (2025). Under Review at NeuroAI workshop in AAAI 2026.

Neuro-AIFactor OrthogonalizationPattern Separation

Microglial NLRP3 Inflammasome Activation and Exosomal miRNA-Mediated α-Synuclein Propagation

In Preparation

A Multi-Omic Strategy to Predict Adaptive DBS Outcomes in Parkinson’s Disease. Neurotechnology Capstone Project.

NeurotechnologyParkinson's DiseaseMulti-Omics

Interpreting DG-Style Pattern Separation Through Embedding Bases

In Preparation

Recovering Class Prototypes as Sparse Linear Combinations of Semantic Embeddings. Computer Science Honors Thesis Project.

Pattern SeparationEmbeddingsHonors Thesis

Regeneration of Underwater Diffusion Models

In Preparation

Using Diffusion to Generate Underwater Images Using SONAR Components and Text Descriptions. Being prepared for submission.

Diffusion ModelsUnderwater ImagingSONAR

02. Experience

Professional journey

Professional Experience

Marriott International

Software Research Engineer (Flex), Loyalty Engineering

June 2024 - December 2024
•
Bethesda, MD
  • ▸Streamlined productivity of the Loyalty IT team through a Jira Ticket assistant, running searches on past Jira tickets and providing recommendations to solve tickets, overall increasing the productivity of the team by 20% and reducing ticket resolution time by 50%
  • ▸Collaborated with Viam and Marriott Design Labs to build a Rover for improving Waste Management in Marriott offices and hotels through auto circulation
AI/MLSoftware EngineeringRoboticsNLP

Apex Fund

Analyst, Quantitative

August 2023 - Present
•
College Park, MD
  • ▸As part of the Apex Fund’s Quantitative division, I started helping out develop strategies, build infrastructure to eventually leading my own desk and strategy with 4 junior analysts
  • ▸Worked on a strategy to trade Brent/WTI futures on Soccer trades made by owners of clubs with significant business in the Oil Production industry
  • ▸Solely responsible for building the infrastructure of the club, from a database to an event driven Backtester to ease development of all future strategies
  • ▸Developed a Dentate Gyrus inspired sparse orthogonalization model for financial factor disentanglement, integrating nonlinear sparse expansion with Hopfield-style attractor dynamics to reduce cross-factor correlation while preserving local geometry, to be used in our portfolio risk assessment models, submitted to NeuroAI workshop at AAAI-2026
Quantitative FinanceMachine LearningStrategy DevelopmentInfrastructure

Teaching Experience

University of Maryland, College Park

Department of Computer Science

Undergraduate Teaching Assistant - CMSC 216 - Introduction to Computer Systems

January 2026 - May 2026
•
College Park, MD
  • ▸Pre-emptive offer from Dr. Christopher Kauffman to be a part of his teaching staff for CMSC 216, Introduction to Computer Systems for Spring 2026
TeachingComputer SystemsC Programming

University of Maryland, College Park

Department of Computer Science

Undergraduate Teaching Assistant - CMSC 216 - Introduction to Computer Systems

August 2023 - May 2024
•
College Park, MD
  • ▸Worked under Dr. Ilchul Yoon and Dr. Christopher Kauffman to teach CMSC 216, Introduction to Computer Systems
  • ▸Handled over 1000 students and taught 5 sections of 40 people each, and graded assignments and exams for students
  • ▸Was voted one of the best TAs from a Staff size of 40 under Dr. Kauffman
TeachingComputer SystemsC ProgrammingLinux

03. Projects & Hackathons

Award-winning innovations

Solyd (Clinical Intelligence Platform)

HackMIT 2025 - Winner 1st Place (Rox)

Engineered an AI pipeline to convert unstructured medical text into a queryable knowledge graph for real-time clinical decision support. Built the agent-mapping subsystem.

AI PipelineKnowledge GraphHealthcare

ThoughtWheels (BCI-Controlled Wheelchair)

TreeHacks 2024 - Top 10 Finalist

Developed a Brain-Computer Interface (BCI) using Muse 2 EEG signals to control a robotic wheelchair. Designed Python-based signal processing pipelines to decode micro-movements and motor imagery from raw EEG data for precise navigation control.

BCISignal ProcessingRobotics

Blindspot (Assistive Computer Vision)

Bitcamp 2023 - Winner 1st Place (Overall)

Developed assistive glasses for the visually impaired utilizing optimized YOLO-based object detection and low-latency embedded vision algorithms.

Computer VisionYOLOEmbedded Systems

MediChain (Decentralized Health Data)

HackMIT 2024 - Winner Best Blockchain Hack

Blockchain network with decentralized storage to securely and efficiently share patient data

BlockchainHealthcareDecentralized Storage

05. Get in Touch

Let's collaborate

I'm actively seeking PhD opportunities in Neuro-AI and Computational Neuroscience for Fall 2026.

Whether you're interested in discussing research collaborations, PhD opportunities, or just want to connect, I'd love to hear from you.

Connect

GitHubkushalkapoor25
LinkedInkushalkapoor
Emailkushalk@umd.edu