RESUME
EDUCATION
B.Sc Computer Science: Data Science | GPA: 3.81
- Courses: Machine Learning, Artificial Intelligence, Bayesian Inference, Software Engineering, Financial Modeling, Marketing and Product Analysis, Economics, Data Structures and Algorithms, Statistics, Calculus I and II, Linear Algebra.
EXPERIENCE
TaxGPT (Research and Engineering Team)
Research Scientist
- 4th engineering hire; designed and built TaxGPT's scalable AI tax-research agent in a SOC 2–compliant production environment, driving 30× growth in weekly active tax-professional users.
- Architected core agentic services, orchestration engines, and tax-law corpus retrieval systems with MCP-based tool and data interfaces, enabling modular retrieval, reasoning, and execution workflows and reducing time to deploy new agents.
- Developed Agent Andrew, an autonomous tax reviewer analyzing up to 500-page filings to flag issues in tax returns.
- Built an evaluation framework with benchmarking, regression tests, and prompt management to compare models, agent architectures, and retrieval approaches, maintaining performance while enabling safe experimentation.
Womp Labs (Research Team)
Machine Learning Intern
- Researched model-agnostic data attribution for CV and NLP, benchmarking interpretability methods and identifying high-impact training data points using NTK-based approximations and LoRA-efficient influence estimation.
- Trained ResNet-18/34/50 on ImageNet-1k and GPT-2 on FineWeb, optimizing 8×A100 FFCV pipelines for high-throughput experimentation.
Universidad Abierta Interamericana (CAETI)
Machine Learning Research Intern
- Researched rule-extraction methods for interpretable deep learning, benchmarking the RxREN algorithm across CV datasets and architectures and achieving 96% fidelity in converting black-box models into symbolic rule systems, enabling transparent auditing of model decision rules for reliability and compliance.
- Performed systematic hyperparameter and architecture studies to strengthen explainability robustness across models.
World Wildlife Fund (Data and Technology Team)
Data Science Intern
- Built Django REST APIs and Azure CI/CD workflows to automate partner data ingestion and standardize unstructured geospatial and tabular inputs into unified database schemas.
PROJECTS
StarcAI (Co-founder)
Built sentiment modeling pipelines over MD&A disclosures in 10-K/10-Q filings to produce risk-calibrated signals guiding the structure and tone of corporate investor communications.
Developed sentiment-conditioned text generation to produce tone-aligned investor communication drafts.
Implemented the first hybrid of Grouped-Query Attention and Mixture-of-Depths in a compact language model, achieving 70% faster training with ~10% accuracy tradeoff compared to baseline architectures.