Hello! I’m Jinxiang Ma — a recent graduate of the University of Michigan, where I earned my Master’s in Industrial and Operations Engineering. I also hold dual bachelor’s degrees in Mathematics and Statistics & Data Science from UC Santa Barbara. My academic and professional path lies at the intersection of data science, data engineering, optimization, and decision science, and I’m passionate about building robust data systems that support real-world decisions in infrastructure, healthcare, manufacturing, and beyond.

With a solid foundation in statistical modeling, algorithm design, and systems thinking, I’ve always been drawn to solving ambiguous problems — the kind where data is messy, business logic is fragmented, and solutions need to balance usability, accuracy, and operational fit.

Professional Experience

#

#

One of the most formative chapters in my professional journey was my two-year internship at the San Francisco Public Utilities Commission (SFPUC). I began as a data science intern and was later promoted to student intern lead. I noticed that project data was tracked in inconsistent Excel files across departments, causing major friction in year-over-year planning and reporting. I led the design and implementation of a new system built on Power Apps and SharePoint, where I developed a user-friendly frontend for project managers and automated data ingestion pipelines using Pentaho Kettle and Python.

To resolve mismatched project identifiers, I built a NLP-based matching pipeline using a hybrid approach of TF-IDF, BM25, and transformer-based embeddings. I also designed QA/QC checks using the SharePoint API to validate data before loading into the agency’s Oracle + Unifier backend system. I built interactive Power BI dashboards that reported budget trends, schedule status, and phase distribution — tools that are now used agency-wide during CIP planning cycles. This end-to-end solution earned me an Employee Recognition Award and taught me how to deliver value through technical innovation grounded in stakeholder workflows.

At Michigan Medicine, I worked on a research project aimed at improving the accuracy and throughput of sterile compounding in hospital pharmacies. After discovering inefficiencies in reconstitution and quality control processes, we proposed a new ultrasound sonication step and modeled the preparation workflow as a hybrid flow shop scheduling problem, incorporating both serial and parallel tasks. I formulated the problem as a mixed-integer program (MIP) in Gurobi and also implemented a discrete-event simulation version using Python to validate feasibility under varying workloads.

To support workload planning, I also trained a SARIMA model on historical preparation data to predict daily compounding volume, which helped pharmacy managers allocate staffing resources more effectively. This experience gave me firsthand exposure to decision science in a healthcare environment — blending optimization, forecasting, and process design in a mission-critical setting.

Earlier in my career, I served as a data analyst at GoGaucho, a UCSB student-run app serving over 10,000 users. There, I analyzed course and review data to optimize in-app recommendations and refine UX design. As a student assistant at the DREAM Lab, I supported data science education by building hands-on tutorials and instructional materials on topics such as web scraping, text analysis, and Python automation. These early experiences sparked my interest in bridging the gap between users, systems, and data products.

Research & Development

#

#

In addition to my applied work, I’ve conducted research at the SOCR Lab on scalable analytics for high-dimensional biomedical data. My focus was on implementing efficient subsampling strategies like reservoir sampling and chunk-based stratification, allowing us to analyze 40GB+ datasets without exceeding memory constraints. I used the SLURM workload manager on Michigan’s HPC cluster to parallelize tasks, and trained SuperLearner ensemble models for feature selection under computational limits. The project emphasized not only statistical robustness, but also the importance of infrastructure-aware data science.

Across all my experiences, I’ve found that I’m most fulfilled when building solutions that go beyond the model — solutions that are scalable, explainable, and directly used by decision-makers. Whether in the public sector, healthcare operations, or cloud analytics, I enjoy navigating the entire data lifecycle: from defining the problem, to building the infrastructure, to delivering dashboards, models, and insights that actually move the needle.

Thank You

#

#

Thank you for visiting my portfolio site. My professional interests span data science, cloud-scale data engineering, reliability analytics, and revenue management — all tied together by a passion for using data to improve how decisions are made and systems are run. I hope you enjoy browsing through my work, and please feel free to reach out if you’d like to connect, collaborate, or just chat about data and systems.

Best regards,
Jinxiang Ma