02 / SELECTED WORK

Real problems.
Working solutions.

Projects, internship work, and research across full-stack systems and applied machine learning. Explore each build for its implementation and results.

Showing 5 entries

Illustration of an HPC job monitoring interface
PROJECT · 202601

HPC Copilot

Built a full-stack HPC platform for job submission, monitoring, and real-time status visualization, supporting 100+ concurrent workflows.

40% lessjob-tracking time
Inside the build
  • Built the platform with FastAPI, React, TypeScript, PostgreSQL, and Docker.
  • Developed an AI assistant for natural-language HPC queries and knowledge retrieval using Dockerized FastAPI services and PostgreSQL.
  • Reduced troubleshooting time by 30% for containerized workflows.
FastAPIReactTypeScriptPostgreSQLDocker
Illustration of a machine learning network
SRL LABS · RESEARCH · 2025–202602

Human Activity Classification

Built a YOLO-based computer vision pipeline processing 1,000+ videos for human activity classification and behavioral analysis.

92%training accuracy with Bidirectional LSTM models
Inside the build
  • Engineered temporal and spatial pose features for sequence modeling.
  • Trained Bidirectional LSTM models with TensorFlow and Keras. The 92% result is training accuracy.
YOLOPythonTensorFlowKerasBidirectional LSTM
Illustration of network traffic flowing through a classification model
SRL LABS · RESEARCH · 2025–202603

Network Intrusion Detection

Developed a TensorFlow/Keras intrusion detection system classifying 14 network traffic categories from CICIDS2017 across benign and malicious traffic.

89%overall classification accuracy
Inside the build
  • Applied dropout and L2 regularization across 78 traffic features to improve generalization on highly imbalanced attack classes.
  • Built preprocessing and evaluation pipelines for precision, recall, F1-score, ROC-AUC, anomaly detection, and inference analysis.
  • Analyzed DDoS, DoS, PortScan, brute-force, web attacks, and benign traffic.
TensorFlowKerasCICIDS2017Python
Illustration of inventory records with role-based access
R3S TRADING · INTERNSHIP · 202504

Inventory Management System

Built a secure full-stack inventory system with role-based access control and JWT authentication, enforcing 100% role coverage.

~85%test coverage in Spring Boot / Spring Security services
Inside the build
  • Reduced security-related defects by approximately 40%.
  • Containerized services with Docker, reducing setup time by approximately 70% and deployment time by approximately 50%.
  • Resolved 95% of critical defects within SLA through defect tracking, root-cause analysis, and validation testing.
Spring BootSpring SecurityDockerRBACJWT
Illustration of articles flowing into a sentiment analysis model
PROJECT · 202505

News Sentiment Analyzer

Built an AI platform with TensorFlow and FastAPI to classify sentiment and detect anomalies across 300+ news articles.

300+news articles analyzed
Inside the build
  • Developed automated data pipelines for article processing and prediction.
  • Built JWT-secured REST APIs and SQLAlchemy persistence.
  • Created model validation workflows using FastAPI, React, and TensorFlow.
FastAPIReactTensorFlowSQLAlchemyJWT

THE COMPLETE PICTURE

Experience,
on one page.

Download resume