Aditya Shrestha

B.E. in Computer Engineering · Kathmandu University, Nepal
AWS Certified Solutions Architect – Associate

I am an undergraduate researcher focusing on self-supervised representation learning for real-world time-series and sensor data. My primary interests include environmental early-warning systems, real-time ML pipeline infrastructure, and topological anomaly detection in transaction networks.

My work is built around pre-registered falsification protocols, transparent reporting of negative results, and lightweight, reproducible software engineering.

Selected Research

Self-Supervised Flood Precursor Detection on USGS Streamgage Records
A temporal encoder trained on public streamgage records to identify precursor signals of flooding prior to threshold exceedance.
[Project Details] [Abstract]
Backpressure-Driven Elastic Feature Windows for Real-Time ML Pipelines
An adaptive feature-windowing runtime designed to bound per-batch latency under variable ingestion load, evaluated against baseline streaming pipelines under a fixed test protocol.
[Project Details] [Abstract]
Self-Supervised Anomaly Detection for Glacial Lake Outburst Flood Precursors via Satellite Time Series
A time-series masked autoencoder pipeline evaluated under a pre-registered falsification protocol, including documented negative findings on the South Lhonak retrospective evaluation.
[Project Details] [Abstract]

Featured Engineering Projects

BankSentinel
A five-agent AI architecture for real-time intrusion detection and behavioral transaction inspection.
[Overview]