Michelle Yang founded DatrixLab to help investment teams decide whether AI tools and data products will hold up in their own research process. She has spent more than a decade using data in investment work, evaluating third-party products, and building the systems that researchers rely on. That range of experience lets her examine both the investment use case and the technical work behind it.
Michelle began her career at Citi in agency mortgage trading and US Credit Strategy. She built research data and tools for people making investment decisions. The work taught her to begin with the question an analyst is trying to answer, then trace the data and assumptions behind the answer. An output can be technically sound and still be of little use if it does not match the team’s definitions or way of working.
At Point72, Michelle worked in data quality and Healthcare Data Insights. She evaluated third-party data products, developed quality controls, and worked with researchers, data scientists, and engineers throughout the data pipeline. A vendor’s description of a dataset was only the starting point. Coverage, methodology, revisions, joins, and delivery reliability all had to be understood before researchers could depend on it.
This was also where Michelle gained extensive experience assessing what vendors offered against what investment teams needed. Product samples and demonstrations tend to show favorable conditions. Actual research introduces firm-specific definitions, imperfect source data, existing systems, and deadlines. A useful evaluation has to account for those conditions.
Michelle later joined Woodline Partners, where she founded and developed the Consumer & Healthcare data team. She built research workflows and alternative-data infrastructure covering ETL, governance, visualization, predictive modeling, and pipeline standardization. The role required close work across investing, data science, and engineering. It also required decisions about which processes should be standardized, which checks needed human review, and how much implementation effort a new source justified.
By then, she had seen the same purchasing problem as a researcher, evaluator, builder, and team leader. Investment teams often have limited time to test a product before signing a contract. Comparing vendor claims is difficult when each demonstration uses different examples and assumptions. Even a capable internal data team may not have room to reproduce a research workflow solely for a buying decision.
Michelle established DatrixLab to do that specific work. She works directly with a client to define the workflow and evaluation criteria, test the product with approved data under realistic conditions, and write a recommendation supported by the findings. The assignment has a defined end. DatrixLab does not take permanent ownership of the client’s technology, provide managed IT, or replace an internal data team.
Her technical experience includes Python, SQL, Snowflake, Databricks, Airflow, and related data systems. She uses that background to inspect data requirements, understand implementation constraints, and work with both investment and technical staff. The aim is straightforward: determine whether the product fits the client’s research process and state the conditions and limitations clearly.
Michelle holds a master’s degree in financial engineering from the University of Michigan and a degree in risk management science from the Chinese University of Hong Kong.