Master of Science in Artificial Intelligence GPA: 3.86/4.00

University of Houston - Downtown | Houston, TX | Graduation: Dec 2024

• Thesis: “Deep Sleep AI: Automating Sleep Stage Discovery from Raw Single-Channel EEG Signals”

• Selected to mentor AI competition participants at GatorByte’24

• Interests: High-Performance Computing, Signal Processing, Bayesian Networks, Robotics, Computer Vision, NLP, AI Ethics

Bachelor of Science in Data Science (Summa Cum Laude) GPA: 3.86/4.00

University of Houston - Downtown | Houston, TX | Graduation: Aug 2024

• Specialization: Computer Science, Applied Statistics, Computational Mathematics

Lead AI Engineer
ABI LLC. | Houston, TX | Dates: May 2022 - Present

• Developed real-time computer vision algorithms for object detection & classification, optimizing botanical quality assessment systems with deep learning models in TensorFlow & PyTorch.

• Engineered AI-driven image processing pipelines using PIL, integrating Convolutional Neural Networks (CNNs), Fourier Transforms (FFTs), and hyperparameter tuning, reducing model loss to 13% in preliminary testing for sustainable agriculture applications.

• Optimized real-time AI inference architectures, enhancing scalability, computational efficiency, & automation in large-scale agricultural product grading systems.

• Collaborated with the Principal Engineer / Technical Co-Founder to refine deep learning model architectures, advancing automated plant quality classification through state-of-the-art CNN-based object recognition.

Student Research Assistant (Genomics)
University of Houston - Downtown | Houston, TX | Dates: Sept 2022 - Dec 2023

• Conducted NSF grant-funded AI research in computational genomics, integrating machine learning, probabilistic modeling, & graph theory to enhance genomic sequence integrity.

• Developed a genotype correction framework using probabilistic anomaly detection, reducing sequencing error rates & achieving 97% accuracy in data restoration.

• Collaborated on a kNN-based scalable data imputation pipeline leveraging machine learning-based missing value prediction, automating genomic preprocessing with feature engineering techniques.

• Engineered graph-based machine learning in NetworkX, implementing community detection algorithms to classify bipartite relationships between long non-coding RNA (lncRNA) and Topologically Associating Domains (TADs), advancing genomic clustering methodologies.

• Benchmarked AI-driven sequence correction algorithms against traditional genomic techniques, improving structural interpretation of genome folding patterns & contributing to computational biology research.

🧠 DeepSleepAI – EEG Signal Automated Classification

Aug 2024 - Present | GitHub Repo: deepsleepai

• Developed PyEEG-GPU, a GPU-accelerated Python library for electroencephalography (EEG) feature extraction, achieving a 10x speedup over baseline methods for real-time signal processing. Designed a multi-stage feature engineering pipeline with CUDA, CuPy, & parallel processing to extract 20+ features across time (variance, skewness, kurtosis), frequency (Power Spectral Density, Hjorth Parameters), & complexity measures (Fractal Dimension, Spectral Entropy, Fisher Information). Integrated this high-performance library into an I/O pipeline, enabling scalable EEG signal analysis for medical & neuroscience applications. Repo: pyeeg-gpu

• Optimized probabilistic learning for sequential EEG data using Transformers, encoding conditional probabilities on a Dynamic Bayesian Network (DBN) from Pgmpy to model adaptive sleep transitions. Improved classification performance on the Sleep-EDF dataset, outperforming CNN-based sleep staging methods with a 5% increase in accuracy & 20-30% reduction in inference time (PyTorch, TensorFlow).

• Designed a modular HPC-driven inference framework, integrating parallel computing (NumPy, SciPy, OpenMP) & containerized deployment (Docker, Linux, PowerShell) for scalable, reproducible AI-driven EEG classification, with a future research focus on Quantum Machine Learning (QML) for EEG signal representation.

• Transcript in progress for journal publication in 2025.

🗣️ Natural Language Processing: LLM & Speech Synthesis Engineering

Aug 2023 - Dec 2023 | GitHub Repo: lang-learn

• Designed & optimized Transformer-based pipelines for medical diagnostics & conversational AI using GPT, Falcon-7B, & BERT, integrating PyTorch, TensorFlow, & Hugging Face frameworks.

• Implemented Quantized Low-Rank Adapters (QLoRA) to reduce Falcon-7B’s memory footprint by 45% while maintaining accuracy for scalable LLM deployment.

• Improved medical query resolution by 18% through weighted loss optimization & custom tokenization strategies for semantic alignment.

• Engineered a CNN-based speech synthesis module by integrating BERT embeddings with Tacotron & WaveNet, enhancing real-time human-computer interaction for voice AI systems.

• Researched reinforcement learning-inspired fine-tuning methods to improve response reliability & explored edge deployment of multi-modal LLMs for real-time inference.

👀 Computer Vision: Object Feature Detection & Classification

Aug 2023 - Dec 2023 | 😶‍🌫️ GitHub Repo: edge-detection | ☝️ GitHub Repo: fingerprints

• Engineered AI-based vision systems for facial recognition, object segmentation, & biometric authentication using CNNs, Transfer Learning (ResNet, AlexNet, VGG), & Vision Transformers (ViT). 

• Developed hybrid pipelines to enhance robustness under occlusion, glare, & variable light. 

• Applied multi-stage edge detection (Canny, gradient-based, RL-inspired weighting) & frequency-domain segmentation for high-precision object & contour recognition. 

• Researched GAN-based generative modeling & self-attention mechanisms to explore next-generation feature extraction techniques.

• 😶‍🌫️ Sub-project 'edge-detection' algorithm detected edges, highlights and shadows for feature extraction analysis using FFT & gradient-based methods. Project explained the Fourier Frequency Transformation process and demonstrated an achieved 96% accuracy in real-world object detection tasks.

• ☝️ Sub-project 'fingerprints' algorithm classified real versus fabricated fingerprints from sensor-detected images, comparing performance using Histogram of Oriented Gradients (HoG) and Local Binary Pattern (LBP) classifiers. Improved classification accuracy by 12% over baseline.

📊 Time-Series Forecasting for Wearable Device Sensors

Jun 2023 - Aug 2023 | GitHub Repo: neural-networks

• Designed & optimized deep learning models (RNN, CNN, LSTM, Bidirectional LSTM) for multivariate time-series classification of tri-axial gyroscope & accelerometer data using TensorFlow (Keras) & PyTorch.

• Preprocess raw sensor inputs into 3D tensors with NumPy & HDF5, enabling memory-efficient training & real-time analysis across variable-length sequences.

• Achieved 91% classification accuracy on unseen sensor data via hyperparameter tuning (dropout, batch size, L2 regularization, bidirectionality) & implemented a hybrid CNN-LSTM model, improving precision 7% over baselines.

• Built an end-to-end pipeline for training with optimization methods (early stopping, model checkpointing, sequence padding, cross-validation), evaluating model performance metrics with F1-score, ROC-AUC, & SHAP-based interpretability for real-world deployment.

🧮 Numerical Methods in Science & Engineering

Jan 2023 - Aug 2024 | GitHub Repo: numerical-methods

• Engineered numerical solvers for ODEs & PDEs using matrix factorizations (PLU, QR, SVD) & iterative methods (Gauss-Seidel, Conjugate Gradient, Newton-Raphson) with Python (NumPy, SciPy).

• Implemented forward, central, & backward finite-difference schemes for elliptic, parabolic, & hyperbolic PDEs, benchmarking solver stability & convergence for high-fidelity physical simulations.

• Applied gradient-based & derivative-free optimization (Newton-CG, BFGS, Nelder-Mead) to minimize multivariable systems, enhancing solver precision for physics-based modeling.

• Solved complex boundary value problems (BVPs) with LU decomposition & adaptive mesh refinement, integrating PCA-like eigenvalue decomposition factorizations for dimensionality reduction in scientific AI systems.

📈 Bayesian Inference on Stock Market Predictions

Jan 2022 - May 2022 | GitHub Repo: stocks

• Built Bayesian inference models for real-time stock prediction in R (RStudio, JAGS) using Markov Chain Monte Carlo (MCMC), hierarchical modeling, & probabilistic forecasting, improving accuracy by 18% over traditional models. 

• Engineered a live data pipeline (Yahoo! Finance API), applied Gamma-Poisson & Normal-Normal priors, & enhanced risk modeling with BayesFactor, Bayesian ANOVA, & 95% credible intervals for robust financial decision-making.

🚀 Predicting Missile Targets

Jan 2024 - May 2024 | GitHub Repo: applied-regression

• Developed applied regression models in R using GLMs, Dplyr, & Tidyverse, applying linear, logistic, & polynomial regression to scientific data. 

• Achieved 88% accuracy predicting missile hit/miss rates, optimized models via AIC/BIC, multicollinearity analysis (VIF), & improved robustness with Cook’s Distance, cross-validation & RMSE evaluation (Caret Package).

🧬 Tumor Detection in Medical Images

Aug 2022 - Dec 2022 | GitHub Repo: tumor-classification

• Built an end-to-end tumor classification pipeline in Python using TensorFlow, Keras, & Scikit-Learn, achieving 96% accuracy with a multi-layer neural network. 

• Engineered large-scale medical data pipelines, automated model selection with PyCaret, & benchmarked classical models (SVM, XGBoost, Random Forest), improving classification efficiency by 32% through cross-validation & ensemble strategies.

🛳️ Missing Passengers Data Imputation & Prediction Modeling

Jan 2023 - Aug 2023 | GitHub Repo: titanic

• Built & optimized supervised & unsupervised ML models in Python using Scikit-Learn, GridSearchCV, & Principal Component Analysis (PCA), achieving 91% accuracy with SVMs & 82% with Naïve Bayes. 

• Applied regression, classification, clustering (K-Means, DBSCAN), & developed an early-stopping gradient descent algorithm, reducing training time by 15%, while preserving 95% variance with dimensionality reduction techniques.

🐧 Ask the Penguins: Modeling & Inference in R, SQL, and Power BI

Aug 2022 - Dec 2022 | GitHub Repo: stats-models

• Performed statistical modeling & inference in R, SQL, Power BI, applying multivariate analysis, regression (GLM), & hypothesis testing with Dplyr, Tidyverse, & Stats Package to analyze environmental & demographic trends. 

• Built predictive pipelines with Caret, conducted geospatial analysis (Leaflet, ggplot2), & developed Power BI dashboards for data-driven decision-making.

⚖️ Social Justice & Recidivism Modeling

Jan 2022 - May 2022 | GitHub Repo: TBD

• Developed machine learning models in Python to assess bias in recidivism prediction algorithms, simulating DOJ decision-making processes & comparing predictions to real recidivism data, revealing systemic bias in risk assessment models like COMPAS; applied Decision Trees & k-Means Clusters with statistical hypothesis testing to identify trends in wrongful convictions & sentencing disparities, visualizing insights in Matplotlib, Seaborn, Choropleth Maps.

🌍 Global Population Forecasting & Statistical Modeling

Aug 2021 - Dec 2021 | GitHub Repo: world-population

• Performed large-scale statistical modeling on WHO global population data, applying probabilistic forecasting, Poisson regression, and time-series analysis in R to predict long-term demographic trends, integrating geospatial visualization (Choropleth maps, ggplot2) & automated data pipelines for efficient extraction, transformation, & reporting.

🤖 Robotics: Navigation, Sensors, & Computer Vision

Jan 2024 - May 2024 | GitHub Repo: TBD

• Designed, assembled, and tested two sensor-guided robots, integrating sensor fusion & motion planning algorithms; calibrated & optimized robots using PID motor control tuning & object detection accuracy; simulated SLAM-based robotic motion in CoppeliaSim, engineered real-time face detection & object tracking with OpenCV & Haar cascades in Python, & implemented motor control using Arduino, Raspberry Pi, and Interactive C.