Selected work

Applied computer vision and AI projects from research to deployment.

These projects map directly to my resume: detection, tracking, media forensics, dataset curation, and ML systems built for real-world constraints.

Hire me for your team See resume

Baseline first: clean data, clear metrics, reproducible pipelines.
Deployment-minded: edge constraints, retrieval systems, and evaluation baked in.
Documentation: concise READMEs, results tables, and handoff notes.

Applied research projects

Work spanning computer vision, media forensics, and edge deployments.

AMC, GE HealthCare, UAlbany - Mar 2023 to Present

Pose estimation, seizure state prediction, and behavior analysis for induced and natural conditions.

Computer Vision Pose Estimation Biomedical

Kitware & UAlbany - Jan 2020 to Dec 2023

Detected falsified media/news using semantic forensics across object, scene text, and human pose signals.

Media Forensics Detection ML

GE Research & UAlbany - Jul 2019 to Dec 2021

Rail yard management using CV analytics: object detection, tracking, and re-identification with edge plus control center architecture.

Detection Tracking Edge

GE Research & UAlbany - Jan 2020 to Jan 2021

Enhanced rail crossing safety using instance segmentation, multi-object tracking, and rule-based logic.

Instance Segmentation Tracking Safety

UAlbany - Jan 2021 to Aug 2021

Analyzed submissions for detection, tracking, re-identification, and road anomaly tasks; reported system-level insights.

Benchmarking Detection Tracking

Research Foundation of SUNY - Jul 2019 to Dec 2019

Added vehicle colors and subclasses and verified ground truth for large-scale traffic analytics.

Dataset Curation Annotation Traffic

Hands-on builds that explore detection, prediction, web, and IoT systems.

Vehicle type and color detection: YOLOv3 and KNN on UA-DETRAC for 12 vehicle types and 9 colors with strong accuracy.
Movie success prediction: Twitter and IMDb mining; ~80% accuracy using ML classifiers.
Stock price prediction: Intraday pipeline using pattern recognition, NLP, storyline analysis across tweets, news, and market data.
Library Management System: Java app handling institutional library workflows.

Social Geo Tagging Web App: PHP app aggregating social data and mapping alumni connections.
Online Yearbook: Full-stack web app for comments and social links.
Smart Attendance System: IoT RFID and Arduino with Google Sheets API backend.
Hand Gesture Control: MATLAB-based webcam gesture recognition for PC control.

The principles I use to move fast without sacrificing quality.

Start with baselines: establish simple, reliable reference points before adding complexity.
Measure and log: track every experiment, metric, and artifact for future comparison.
Communicate early: share progress with concise docs and demos to align stakeholders.
Ship in loops: iterate in small, testable increments that can be deployed and observed.

Start a collaboration Read my notes