Mika Okamoto

I'm a B.S./M.S. student in Computer Science at the Georgia Institute of Technology, graduating in May 2027. I'm passionate about Explainable AI and Automated Reasoning, and I'm especially interested in how these areas can be applied to complex, real-world domains.

I currently work as a Research Assistant at the GTRI ARCAID Lab, where I explore hierarchical representation learning to improve reasoning systems. I also contribute to the GT Financial Services Innovation Lab, developing natural language processing and explainability techniques for financial market analysis. In addition, I collaborate with the Entertainment Intelligence & Human-Centered AI (EI & HCAI) Lab on projects related to explainable AI and machine unlearning.

Beyond research, I'm interested in software engineering, machine learning engineering, and data science. This summer (2025), I’ll be interning as a Software Engineer at Two Sigma Investments. I'm always happy to connect — feel free to reach out!

Experience Highlights

View details for Two Sigma Investments experience
June 2025 - Aug 2025

Two Sigma Investments

Software Engineering Intern

Upcoming, Summer 2025.

View details for Georgia Tech Research Institute experience
Jan 2024 - Present

Georgia Tech Research Institute

Advisors: Clayton Kerce, Pat Langley

Automated reasoning through learning hierarchical representations.
View details for Raytheon Technologies experience
May 2024 - Aug 2024

Raytheon Technologies

AI Algorithm Engineering Intern

Integrated explainable AI methods into data workflows.

View details for GT FinTech Lab (FSIL) experience
Jan 2024 - Present

GT FinTech Lab (FSIL)

Advisor: Sudheer Chava

Natural language processing and explainable AI for financial markets.
View details for UCSD Palmer Lab experience
June 2023 - Sept 2023

UCSD Palmer Lab

Advisor: Abraham Palmer

Applied machine learning for genetic trait prediction.

View details for GT EI & HCAI Lab experience
Jan 2025 - Present

GT EI & HCAI Lab

Advisor: Mark Riedl

Explainable AI and machine unlearning.

Research & Publications

Highlighted Research Projects

BELLA: Budget-Efficient LLM Selection via Automated Skill Profiling

Mika Okamoto, Ansel Erol, Glenn Matlin, Sudheer Chava

BELLA addresses the challenge of selecting the most suitable Large Language Model (LLM) for a given task under real-world constraints (latency, budget). BELLA analyzes LLM outputs to identify interpretable skills and weaknesses, creating structured profiles to recommend models that offer the best utility within user defined resource constraints, bridging the gap between benchmark accuracy and deployment needs.

Paper Poster (MLSys YPS '25)

RATTACA: Genetic predictions in Heterogeneous Stock rats offer a new tool for genetic correlation and experimental design

Benjamin B Johnson, Thiago M Sanches, Mika H Okamoto, Khai-Min Nguyen, Clara A Ortez, Oksana Polesskaya, Abraham A Palmer

RATTACA introduces a novel research paradigm using genetic predictions from Heterogenous Stock (HS) rats (via linear mixed models) to efficiently identify putative genetic correlations and facilitate a priori sampling of individuals with extreme trait values, offering a viable alternative to intensive experimental phenotyping.

Preprint Poster Code

Publications

*= denotes equal contribution

FLaME (Holistic Finance Language Model Evaluation). Glenn Matlin*, Mika Okamoto*, Huzaifa Pardawala*, Yang Yang, Sudheer Chava. Accepted to ACL Findings 2025. Paper Demo Code
Enhancing Military Family Readiness and Resilience Programs Through LLM-Generated Synthetic Data and Natural Language Interfaces. Anantharaman Iyer*, Mika Okamoto*, Agustin Lorenzo, Blair Johnson, Moon Kim, Clayton Kerce, Margarita Gonzalez. Accepted to MHSRS 2025.
BELLA: Budget-Efficient LLM Selection via Automated Skill Profiling. Mika Okamoto, Ansel Erol, Glenn Matlin, Sudheer Chava. Poster presentation at MLSys Young Professionals Symposium (YPS) 2025. Preprint Poster PDF
RATTACA: Genetic predictions in Heterogeneous Stock rats offer a new tool for genetic correlation and experimental design. Benjamin B Johnson, Thiago M Sanches, Mika H Okamoto, Khai-Min Nguyen, Clara A Ortez, Oksana Polesskaya, Abraham A Palmer. Preprint Code
Benchmarking LLM Performance on Financial Tasks with Optimized Prompting Strategies. Mika Okamoto. Paper
Genome-Wide Association Study of Age-Related Hearing Loss in CFW Mice Identifies Multiple Genes and Loci, Including Prkag2. Oksana Polesskaya, Ely Boussaty, Riyan Cheng, Olivia A. Lamonte, Thomas Y. Zhou, Eric Du, Thiago Missfeldt Sanches, Mika Okamoto, Abraham A Palmer, Rick Friedman. Journal of the Association for Research in Otolaryngology. Paper

Projects & Work Experiences

Work Experience

Raytheon Technologies (RTX)

AI Algorithm Engineering Intern | May 2024 – August 2024

Developed a Python framework to integrate explainable AI techniques into data workflows, enhancing customer communication and trust.
Integrated MLflow to improve MLOps tracking and reproducibility, saving 50+ hours/month.
Created a multithreaded security system that improved real-time data processing speed by 5x for object detection, integrating 5 cameras and 3 ML models on limited computational resources.

Research Lab Experiences

Georgia Tech Research Institute (GTRI) - ARCAID

Research Assistant | January 2024 – Present

Advisor: Clayton Kerce Advisor: Pat Langley

Investigating scientific claim decomposition and verification using AI agents and search algorithms.
Researching techniques for learning generalized methods and solutions to planning problems via hierarchical problem networks to reduce search complexity.
Built a RAG-enhanced chatbot and simulation pipeline (code generation, execution, debugging) to accelerate research onboarding and experimentation.

Georgia Tech - Financial Services Innovations Lab (FinTech Lab)

Research Assistant | January 2024 – Present

Advisor: Sudheer Chava BELLA (MLSys YPS 2025) LLM Prompting Report

Researched budget-efficient LLM selection via automated, interpretable skill profiling.
Created a finance holistic evaluation benchmark for assessing language model capabilities.
Benchmarked LLM prompting methods for financial sentiment classification, achieving 10% relative improvement over baselines via automated prompt optimization.

Georgia Tech - EI & HCAI Lab

Research Assistant | January 2025 – Present

Advisor: Mark Riedl

Researching machine unlearning methods to identify training data correlations with language model performance on specific task domains.

Palmer Lab at UCSD

Bioinformatics Intern | June 2023 – October 2023

Advisor: Abraham Palmer RATTACA Mice Hearing Loss

Improved phenotype trait prediction machine learning pipeline with rat genetic data through data reduction (7.3M features to 50k features) with minimal performance degradation.
Performed data analysis on phenotype datasets for quality control and GWAS pipeline preparation.

Mika Okamoto

Recent News & Updates

Experience Highlights

Two Sigma Investments

Georgia Tech Research Institute

Raytheon Technologies

GT FinTech Lab (FSIL)

UCSD Palmer Lab

GT EI & HCAI Lab