Skip to content
Research | ML | Benchmarking

Research built for real engineering use.

I run reproducible experiments in ML and optimization, then translate findings into practical systems.

Reinforcement Learning Music ML Algorithm Analysis

Method

Reproducible runs

Metrics

Tracked and comparable

Validation

Baseline + ablations

Output

Code + report ready

Each study favors repeatability over one-off results.

Focus Areas

Music ML

SVM key classification with chroma, MFCC, and spectral features.

SVMFeature engineeringAudio analysis

Deep RL

DQN and PPO on LunarLander with replay and curriculum strategies.

DQNPPOReward shaping

Optimization

Hyperparameter tuning and benchmark-driven model comparison.

TuningBenchmarkingComparative analysis

Selected Research

Music Key Detection with SVM

2024

Built a 24-key classifier with PCA and tuning. Final accuracy: 67.12%.

View repo

Advanced RL on LunarLander

2024

Compared DQN and PPO with replay, shaping, and multi-seed evaluation.

View repo