01
In ProgressFeatured
2025AI-Powered Portfolio Assistant (RAG-based System)
LLM-powered assistant using RAG to enable interactive exploration of my projects and experience.
LLMRAGNLPEmbeddingsVector DB+3
Work & Projects
Projects
A collection of my work in machine learning, security research, and educational tools.
8 / 8 projects
02
CompletedFeatured
2024Personalized Movie Recommendation System
End-to-end recommendation engine using collaborative filtering, matrix factorization, and neural methods.
Pythonscikit-learnPyTorchFlaskPostgreSQL+2
03
CompletedFeatured
2025Skynet – AQI Prediction System
ML pipeline to forecast Air Quality Index using NASA TEMPO, OpenAQ, weather, and traffic data.
Pythonscikit-learnpandasNumPyML Pipeline+1
04
In ProgressFeatured
2025R2D2 – Experimental Transformer-based LLM
Building transformer architectures from scratch to understand LLM internals and training dynamics.
PythonPyTorchTransformersNLP
05
CompletedFeatured
2025NeuralLog – Intelligent Activity Tracking System
Personal ML system to track daily activities, analyze behavior patterns, and generate insights.
PythonMachine LearningData Analysis
06
Completed
2021Automatic License Plate Recognition (ALPR)
Computer vision system for vehicle license plate detection and text extraction.
OpenCVTesseractPythonOCR
07
Completed
2021COVID-19 Safeguard System
Real-time monitoring system using computer vision for safety compliance.
Computer VisionTensorFlowOpenCV
08
Completed
2020Research & Experimentation
Experimentation & Case Studies
Hands-on experiments, security research, and applied ML case studies — work that lives at the boundary of exploration and engineering.
AI Security Research · Clemson University2024
Adversarial Attacks – Experimentation & Case Study
- Studied data poisoning, adversarial patches, and evasion attacks on computer vision models.
- Conducted experiments on monocular depth estimation (MDE) systems in autonomous vehicles.
- Analyzed impact on scene understanding, perception, and depth estimation pipelines.
- Evaluated attack transferability across models and threat settings.
Computer VisionAdversarial MLDeep LearningPythonAI Security
AI Security Research · Clemson UniversitySept 2024 – Dec 2024
LLM Defense Evaluation Framework
- Designed automated testing pipelines for probing LLM responses.
- Evaluated token-level patterns, semantic behavior, and refusal mechanisms.
- Measured latency, response consistency, and attack success rates.
- Benchmarked multiple LLMs under diverse jailbreak strategies.
LLMPythonEvaluationAI SecurityAdversarial AI