• Enhanced the accuracy and reliability of LLM models on math and computer-science tasks for high-profile client projects, improving user satisfaction by refining and optimizing model responses using human feedback (RLHF).• Reduced error rates in client deliverables by reviewing other employee work as a Senior Reviewer, contributing to faster project turnaround and improved team efficiency.• Improved client outcomes by ensuring that AI-generated prompts aligned precisely with business… Show more • Enhanced the accuracy and reliability of LLM models on math and computer-science tasks for high-profile client projects, improving user satisfaction by refining and optimizing model responses using human feedback (RLHF).• Reduced error rates in client deliverables by reviewing other employee work as a Senior Reviewer, contributing to faster project turnaround and improved team efficiency.• Improved client outcomes by ensuring that AI-generated prompts aligned precisely with business needs, reducing rework and accelerating client project timelines. Show less

• Stabilized a mission-critical anomaly detection system, preventing downtime caused by mis-formatted data and ensuring reliable data flow, directly reducing escalations and improving client trust in AWS services.• Developed a scalable key performance indicator (KPI) observation system in Apache Spark, capable of analyzing terabytes of daily data, which increased anomaly detection accuracy and scalability for future data growth.• Streamlined debugging processes, saving engineers an… Show more • Stabilized a mission-critical anomaly detection system, preventing downtime caused by mis-formatted data and ensuring reliable data flow, directly reducing escalations and improving client trust in AWS services.• Developed a scalable key performance indicator (KPI) observation system in Apache Spark, capable of analyzing terabytes of daily data, which increased anomaly detection accuracy and scalability for future data growth.• Streamlined debugging processes, saving engineers an average of 3.4 hours per event, allowing them to focus on higher-priority tasks and decreasing false positive escalations by 13%. Show less

• Designed and implemented a high-fidelity Python-based simulation model for adolescent suicide, used by experts at the Centers for Disease Control and Prevention (CDC) to inform policy-making and interventions, enhancing the model’s real-world application to public health.• Collected, cleaned, and merged dozens of distinct data sources on adolescent suicide and ideation.• Performed machine learning parameter optimization techniques to ensure model fidelity to real-world suicide and… Show more • Designed and implemented a high-fidelity Python-based simulation model for adolescent suicide, used by experts at the Centers for Disease Control and Prevention (CDC) to inform policy-making and interventions, enhancing the model’s real-world application to public health.• Collected, cleaned, and merged dozens of distinct data sources on adolescent suicide and ideation.• Performed machine learning parameter optimization techniques to ensure model fidelity to real-world suicide and ideation statistics.• Built a containerized, scalable, and maintainable PHP Laravel web platform, enabling CDC experts to run and analyze simulations with reduced processing time and minimal technical barriers.• Facilitated cross-functional collaboration by leading biweekly meetings, ensuring the timely integration of public health, statistics, and technical inputs, which improved the model’s effectiveness.• Delivered model and web platform with all desired features, comprehensive user tutorials and documentation, and sample simulation scenarios.• Publication in CDC Review:"An Artificial Intelligence Approach to Support Youth Suicide Prevention Initiatives in the U.S.A"L. Liang, R. Schuerkamp, K. L. Rice, M. M. Brown, N. Nataraj, J. Mendoza-Alonzo, C. R. Harper, C. Florence, B. D. Maldonado, B. Moran, P. J. Giabbanelli Show less

• Led efforts to incorporate student feedback in route grading system, redesigning it to be a color-based grading range. This change resulted in higher satisfaction from climbers and a measurable decrease in complaints about “sandbagging” climbs.• Set creative bouldering and top-rope routes of various grades on a weekly rotating schedule for college climbing gym.• Created and operated a route-setting technique improvement workshop at the Midwest Outdoor -Leadership Conference to help… Show more • Led efforts to incorporate student feedback in route grading system, redesigning it to be a color-based grading range. This change resulted in higher satisfaction from climbers and a measurable decrease in complaints about “sandbagging” climbs.• Set creative bouldering and top-rope routes of various grades on a weekly rotating schedule for college climbing gym.• Created and operated a route-setting technique improvement workshop at the Midwest Outdoor -Leadership Conference to help climbers and route-setters across Universities in the Midwest to become better route-setters.• Maintained a professional, clean, safe, efficient work environment through climbing hold cleaning shifts, organizing the holds by color and size, and receiving belay certification and CPR/First Aid certification. Show less

• Led a research team to deliver network analysis models to the CDC, improving the agency’s ability to predict adolescent suicide risks and apply effective interventions through data-driven insights.• Analyzed several previous network-based models of adolescent suicide, comparing key network statistics such as node centrality, edge density, and node neighborhoods.• Applied existing CDC adolescent suicide prevention strategies to the models to demonstrate how the model can represent… Show more • Led a research team to deliver network analysis models to the CDC, improving the agency’s ability to predict adolescent suicide risks and apply effective interventions through data-driven insights.• Analyzed several previous network-based models of adolescent suicide, comparing key network statistics such as node centrality, edge density, and node neighborhoods.• Applied existing CDC adolescent suicide prevention strategies to the models to demonstrate how the model can represent policy interventions.• Co-authored a report as first author for the Cambridge Journal of Network Science:"Guiding prevention initiatives by applying network analysis to systems maps of adverse childhood experiences and adolescent suicide"B. D. Maldonado, R. Schuerkamp, C. M. Martin, K. L. Rice, Nisha Nataraj, M. M. Brown, C. R. Harper, C.Florence, P. J. Giabbanelli Show less

• Optimized system performance by implementing a parallel load testing framework, ensuring reliable daily retraining of anomaly detection models and reducing system bottlenecks by 7%.• Researched industry-standard machine-learning model serving platforms, utilizing the parallel load testing system to collect comparative metrics on platform stability, scalability, fault resistance, and cost.• Deployed the most performant endpoint into the existing system.• Rectified long-standing… Show more • Optimized system performance by implementing a parallel load testing framework, ensuring reliable daily retraining of anomaly detection models and reducing system bottlenecks by 7%.• Researched industry-standard machine-learning model serving platforms, utilizing the parallel load testing system to collect comparative metrics on platform stability, scalability, fault resistance, and cost.• Deployed the most performant endpoint into the existing system.• Rectified long-standing problems with anomaly detection model reload times and on-call alarm sensitivity. Show less

• Automated alarm configuration management, reducing manual effort by developers and improving response times across all Amazon.com fulfillment centers, increasing system throughput.• Developed and tested a system that used a central DynamoDB database to maintain on-call alarm configurations through AWS Lambda broadcasting updated configurations across all fulfillment centers.• Accumulated data on the prototype’s success on various key metrics.• Gained leadership approval by… Show more • Automated alarm configuration management, reducing manual effort by developers and improving response times across all Amazon.com fulfillment centers, increasing system throughput.• Developed and tested a system that used a central DynamoDB database to maintain on-call alarm configurations through AWS Lambda broadcasting updated configurations across all fulfillment centers.• Accumulated data on the prototype’s success on various key metrics.• Gained leadership approval by presenting the prototype and influencing the future system designs for the team. Show less

• Mentored group of 23 freshmen students from diverse and disadvantaged backgrounds throughout their first semester.• Guided students through class registration process, navigating living alone, finding their niche, and fully embracing their college experience.• Educated incoming students about important topics in Diversity, Equity, and Inclusion (DEI) such as privilege, intersectionality, allyship, and navigating the world as individuals with complex or marginalized identities.•… Show more • Mentored group of 23 freshmen students from diverse and disadvantaged backgrounds throughout their first semester.• Guided students through class registration process, navigating living alone, finding their niche, and fully embracing their college experience.• Educated incoming students about important topics in Diversity, Equity, and Inclusion (DEI) such as privilege, intersectionality, allyship, and navigating the world as individuals with complex or marginalized identities.• Created/edited videos for the Miami University Center for Student Diversity and Inclusion needs, with notable projects such as the Women of Excellence award presentation videos and the Lavender Graduation celebration video Show less

• Collected and cleaned data about player movements, actions, and decisions for the Miami Women's Field Hockey Team.• Delivered key insights from the data to the coaching staff of the team to help the players improve.

• Conducted research with electro-encephalography (EEG) brainwave data from five Air Force research subjects.• Trained machine learning models with 85.6% accuracy, advancing research in brainwave analysis and setting a benchmark for future studies on Gabor patch perception.• Investigated data underlying model using factorial experiments to determine which data signals resulted in a model and whether hand dominance affected the results, as previous research suggested.• Validated that… Show more • Conducted research with electro-encephalography (EEG) brainwave data from five Air Force research subjects.• Trained machine learning models with 85.6% accuracy, advancing research in brainwave analysis and setting a benchmark for future studies on Gabor patch perception.• Investigated data underlying model using factorial experiments to determine which data signals resulted in a model and whether hand dominance affected the results, as previous research suggested.• Validated that N200 peak amplitude accounted for 96.3% of results, a key finding confirming existing research contributed to the broader scientific understanding of brain activity and perception in neurological research. Show less

• Researched the efficacy of short-cutting costly HIV cellular automata (CA) simulations using pre-trained image machine learning models.• Optimized CA-based HIV models, improving computational efficiency by reducing simulation runtime and enabling larger-scale simulations, thus advancing HIV research capabilities.• Developed ML models to predict end-case HIV simulation outcomes from initial conditions, reducing the need for costly full-length simulations and enabling faster iteration… Show more • Researched the efficacy of short-cutting costly HIV cellular automata (CA) simulations using pre-trained image machine learning models.• Optimized CA-based HIV models, improving computational efficiency by reducing simulation runtime and enabling larger-scale simulations, thus advancing HIV research capabilities.• Developed ML models to predict end-case HIV simulation outcomes from initial conditions, reducing the need for costly full-length simulations and enabling faster iteration for research teams.• Discovered failure point in model optimization called “anisotropic fronts” caused by race conditions on multi-threaded models, spawning a new inquiry of research into verifying simulation accuracy after optimizations.• Unpublished Report:"Predicting HIV Viral Body Load through Image Analysis"J. Li, B. D. Maldonado, G. Skidmore, Z. Xu, D. Calovini Show less

• Streamlined data pipelines by compiling and balancing a large proprietary patient dataset, increasing the speed and accuracy of ADHD diagnosis modeling.• Analyzed trends in data across previously identified user tasks to find possible signals for ADHD diagnosis.• Trained several rounds of models to detect ADHD in sample data, applying industry-standard practices such as hyper-parameter optimization and factorial experiments where appropriate.• Identified gaps in patient vocal data… Show more • Streamlined data pipelines by compiling and balancing a large proprietary patient dataset, increasing the speed and accuracy of ADHD diagnosis modeling.• Analyzed trends in data across previously identified user tasks to find possible signals for ADHD diagnosis.• Trained several rounds of models to detect ADHD in sample data, applying industry-standard practices such as hyper-parameter optimization and factorial experiments where appropriate.• Identified gaps in patient vocal data for ADHD diagnosis and proposed a novel method for data analysis,influencing future research directions at NeuroLex Labs.• Awarded first place in the NeuroLex 2020 Summer Fellowship program for excellence in ADHD research, innovative data modeling techniques, and impactful presentation of findings. Show less

• Researched existing body of information around reinforcement and deep learning.• Completed course in Computer Vision in Python to understand and apply concepts to a physical car• Constructed a small electronic car utilizing the NVIDIA Jetson Nano Developer Kit.• Applied concepts of deep learning and computer vision to teach the car to drive and avoid obstacles.• Demonstrated results and research findings in an end-of-internship presentation to advisor.

• Implemented research codes from pseudocode into Java.• Learned basics of lab research and operation.

• Assisted middle school band directors with introducing new students to playing saxophone, basic music reading/playing, and how to maintain rhythm.• Accelerated the learning of students who were struggling with adopting the new concepts through one-on-one sessions.

• Managed high-intensity environment and temperature of soccer matches involving 22 on-field players, 2-6 coaches on the sideline, and dozens of spectators and substitute players.• Routinely de-escalated situations using decisive and effective techniques to maintain game flow and keep injury rates at a minimum.• Upheld and enforced the Laws of the Game according to FIFA, regularly keeping up to date with yearly revisions and clarifications.• Mentored new referees and attending… Show more • Managed high-intensity environment and temperature of soccer matches involving 22 on-field players, 2-6 coaches on the sideline, and dozens of spectators and substitute players.• Routinely de-escalated situations using decisive and effective techniques to maintain game flow and keep injury rates at a minimum.• Upheld and enforced the Laws of the Game according to FIFA, regularly keeping up to date with yearly revisions and clarifications.• Mentored new referees and attending several referee improvement workshops as a member of the Advanced Mentor Group. Show less