Research Proposals & Grants

Active multi-year grants, sponsored industrial R&D partnerships, and competitive funding proposals focused on graph learning, edge vision, and trustworthy AI.

Active Grants 3 $1.20M Total Funding

Under Review 2 $570K Requested · Q1 2027

Success Rate 42% ↑ +5% vs prior cycle

Completed 8 7 Papers & 4 Models Shipped

Active Multi-Year Grant Role: Co-PI & Lead Architect

Grant ID: NSF-24-401

Federated Learning Frameworks for Privacy-Preserving Healthcare Records

Problem Statement & Objective

Developing robust gradient aggregation algorithms that handle non-IID clinical data distributions across geographically distributed hospital networks while guaranteeing strict $(\epsilon, \delta)$-differential privacy during multi-institutional model training.

Core Methodologies

Federated Averaging Differential Privacy Homomorphic Encryption Medical Imaging

Timeline: Jan 2024 – Dec 2026 Year 3 of 3 · 80% Complete

Funding Agency: NSF / IIS Division
Award Amount: $450,000 USD
Deliverables to Date: 2 Papers, 1 Clinical Benchmark

View Proposal Details

Active Research Fellowship Role: Principal Investigator

Grant ID: BIO-25-118

Attention-Guided Graph Similarity Learning for Multi-Modal Protein Function Discovery

Problem Statement & Objective

Designing hierarchical Graph Convolutional Networks with cross-graph attention mechanisms to compute structural similarity across 3D protein contact maps and sequence embeddings, accelerating automated gene ontology annotation.

Core Methodologies

Graph Neural Networks Graph Similarity Learning Cross-Attention Bioinformatics

Timeline: Mar 2025 – Feb 2027 Phase 2 of 3 · 60% Complete

Funding Agency: Global Computational Biology Consortium
Award Amount: $570,000 USD
Deliverables to Date: 1 IEEE Paper, 1 Manuscript in Prep

View Proposal Details

Active Industry Partnership Role: Lead ML Engineer

Grant ID: IND-25-002

Real-Time Anomaly Detection in High-Throughput Textile Manufacturing

Problem Statement & Objective

Deploying quantized YOLOv8 object localisation paired with an ensemble of lightweight Vision Transformers on embedded edge accelerators to detect micro-defects on continuous textile looms within sub-10ms latency budgets.

Core Methodologies

YOLOv8 & Ensemble CNNs INT8 Quantization Spatial Attention Edge TensorRT

Timeline: Jun 2025 – Dec 2026 Phase 2 · 70% Complete

Industry Partner: Industrial Vision Consortium
Contract Amount: $180,000 USD
Deliverables to Date: ACM Publication & Edge SDK

View Proposal Details

Stage 2 Panel Evaluation Proposal ID: SRI-26-089

Submitted: Aug 15, 2026

Behavioral & Factual Auditing Framework for Autonomous Multi-Agent RAG Systems

A standardized verification suite to quantify hallucination drift, retrieval poisoning resilience, and source attribution fidelity across collaborative CrewAI and LangChain enterprise pipelines.

Retrieval-Augmented Generation Multi-Agent Systems AI Safety & Attribution

Target Agency Trustworthy AI Research Initiative

Requested Budget $320,000 USD (24 Months)

Expected Notification January 2027

Under External Peer Review Proposal ID: IEEE-PES-26-412

Submitted: Sep 28, 2026

Multi-Agent Reinforcement Learning for Decentralized Microgrid Stability Control

Extending prior smart grid stability prediction models into decentralized MARL controllers capable of autonomous load shedding and frequency regulation under high renewable energy penetration.

Multi-Agent RL Smart Grid Topologies Decentralized Control

Target Agency Clean Energy & Grid Resilience Fund

Requested Budget $250,000 USD (18 Months)

Expected Notification March 2027

Successfully Concluded 2023 – 2025 · Award: $140,000 USD

Urban Traffic Flow Prediction via Spatio-Temporal Graph Neural Networks

Formulated dynamic graph edge re-weighting algorithms to forecast regional arterial congestion with a 14% MAE improvement over recurrent baselines, resulting in 2 peer-reviewed conference publications and an open-source benchmark library.

2 IEEE Publications Open-Source PyTorch Package

View Published Findings

Successfully Concluded 2024 – 2025 · Award: $95,000 USD

Thresholded Input Guided UNet++ (TIG-UNet++) for Automated Pulmonary Segmentation

Developed a modified nested U-Net architecture guided by adaptive thresholded inputs for high-precision lung region extraction and tuberculosis screening in noisy chest radiographs. Published at ICTIS 2025 and IEEE ICEEICT.

ICTIS 2025 Paper Clinical Screening Pipeline

View Published Findings

Archived Sponsored Projects & Seed Grants (2021 – 2024)

6 Concluded Initiatives

2024 Hybrid YOLOv8 & Vision Transformer Framework for Automated Threat Detection

Sponsored Security Vision Seed Grant · Output: 1 First-Author Manuscript

$65,000 USD

2024 Early Readmission Risk Minimization for Diabetic Cohorts via Tabular Ensembles

Healthcare Analytics Grant · Output: IEEE Conference Publication

$45,000 USD

2023 Correlation-Based Feature Extraction for High-Dimensional Sensor Classification

UIU Research & Innovation Grant · Output: IEEE ICCIT 2024 First-Author Paper

$30,000 USD

2023 On-Premise Enterprise Knowledge Assistant (OwnGPT Prototype Phase)

Genuine Technology Research R&D Allocation · Output: Production RAG Platform

$55,000 USD

2022 Neural Sequence-to-Sequence Transliteration for Bengali Text-to-IPA Phonetics

Computational Linguistics Micro-Grant · Output: Open-Source Dataset & Model

$18,000 USD

2021 Transfer Learning Benchmarks for Automated Diabetic Retinopathy Grading

Undergraduate Biomedical Vision Fellowship · Output: Diagnostic Web Tool

$15,000 USD