Hi, I'm Aswin P T
AI/ML Engineer & Drone Vision Researcher
Graduate student in Machine Intelligence at Digital University Kerala (DUK). I develop real-time computer vision pipelines for aerial drone surveillance (YOLOv8, CSRNet on Jetson Orin Nano at 42 FPS) and privacy-preserving Federated Graph Neural Networks with Differential Privacy (ε = 0.5).
Bridging Research Rigor with Production Engineering
I am a Machine Intelligence graduate student at the Kerala University of Digital Sciences, Innovation and Technology (DUK) and a Research Assistant at Quantumweave Intelligence. My work focuses on building robust computer vision systems that run in resource-constrained environments and scalable distributed AI frameworks.
At Quantumweave, I engineer end-to-end UAV surveillance pipelines—combining real-time object tracking (YOLOv8 + ByteTrack) with crowd density estimation (CSRNet), optimized for low-latency onboard inference on NVIDIA Jetson Orin Nano hardware.
In parallel, my research delves into Federated Graph Learning, where I design privacy-preserving algorithms enabling multi-institutional fraud detection under strict Differential Privacy budgets without exposing raw records.
Work Experience & Academic Foundation
Research Assistant – AI & Data Science
CURRENTLeading development of real-world drone computer vision systems and aerial surveillance pipelines:
- Architecting end-to-end vision pipelines for real-time crowd counting (CSRNet), vehicle tracking, and suspicious object detection (YOLOv8 + ByteTrack).
- Optimizing deep learning inference pipelines with INT8 TensorRT for execution on onboard low-power edge compute (NVIDIA Jetson Orin Nano at 42 FPS).
- Developing video ingestion, frame deduplication, alert gating, and geospatial coordinate mapping for field drone deployments.
AI/ML Intern
Delivered production-grade clinical AI components and automated medical document intelligence:
- PI-RADS Prostate AI Platform: Built clinical prostate scoring logic, diagnostic image sector handling, and 3D sector mapping modules with FastAPI.
- Shear Wave Elastography Parser: Created a zero-codebase modular medical report extraction pipeline with strict Pydantic contracts and regex normalization.
Master of Science (M.Sc.) in Machine Intelligence
Featured Engineering & Research Projects
Suspicious Object & Intrusion Detection for UAV Surveillance
End-to-end aerial computer vision pipeline with real-time crowd counting and object tracking optimized for drone edge deployments.
FedGraph-Shield: Hierarchical Federated Graph Learning
Two-tier hierarchical federated framework enabling distributed banking institutions to detect fraud rings without exposing customer data.
KuTo: AI-Powered Tour Guide Marketplace
Intelligent travel marketplace connecting travelers with verified local guides via real-time reverse bidding and AI itinerary planning.
Autonomous Generative E-Commerce Storefront Agent
Autonomous agent pipeline transforming natural language prompts into complete, production-ready storefronts and catalog taxonomies.
Dynamic Connectome fMRI Reality Monitoring Predictor
Graph Attention Networks (GATs) modeling resting-state functional brain connectivity to detect cognitive fatigue and reality monitoring failure.
Adaptive Confidence Modeling for 3D Dressed Human Shape
Image-driven confidence estimation layer resolving monocular 3D body estimation errors caused by loose clothing and occlusions.
PETZZO: Comprehensive Pet Care Platform
Full-stack pet healthcare management ecosystem connecting pet parents with hospitals, veterinary clinics, and GPS pet tracking.
Technical Toolkit & Engineering Arsenal
Computer Vision & Edge AI
8 toolsModel architecture, training & fine-tuning
Aerial detection, custom training & quantization
Dilated convolution crowd counting
Multi-object trajectory tracking
Image transformation & spatial filtering
INT8/FP16 quantization & engine optimization
Edge deployment & hardware acceleration
GPS mapping & bounding coordinate transforms
Graph Neural Networks & Federated Learning
5 toolsGCN, GAT, GraphSAGE heterogeneous graphs
Hierarchical federated client-server coordination
RDP composition & Gaussian noise mechanisms
Topological graph metrics & cycle detection
Dynamic connectome & node attribution
Backend, APIs & Cloud Systems
8 toolsAsync microservices, high-throughput endpoints
Primary production & research language
Multi-stage containerization & edge images
Relational schema design & indexing
Low-level optimization & algorithms
Full-stack web services & ORM
Compute instances & storage buckets
Modern performant client interfaces
Let's Build Something Intelligent Together
I am actively open to discussions regarding AI/ML Engineering opportunities, Aerial Computer Vision projects, and research collaborations.
Send a Direct Message
Fill out the form below to initiate correspondence directly with Aswin P T.