Regression, forecasting, statistical testing, and dashboards for small companies and startups with a specific decision to make — not a report that gets filed away.
At Adore Medical, I analyzed over 10,000 production records and built demand forecasting models that a cross-functional team could actually use. Not models that sat in a report. Models that changed how the company planned inventory. In doing so, I learned what textbooks rarely teach: a technically correct result that no one acts on is not a result.
I am a Mathematics and Statistics student at Boston University with a minor in Computer Science. I work in R and Python. I have built regression models, ANOVA frameworks, classification engines, and dashboards across medical manufacturing, public health, and energy datasets. I take on a small number of freelance projects each quarter — every engagement begins with a specific question and ends with a deliverable your team can use.
57% of high-performing companies alter long-term strategy based on data. Most still lack the analytical infrastructure to act on what they already have. — McKinsey
Each engagement begins with a specific question. It ends with a deliverable your team can act on.
A structured audit of your dataset — distributions, anomalies, relationships, gaps — with a written interpretation designed for non-technical stakeholders. You will know what your data contains and what questions it can and cannot answer.
A validated predictive model built to your question — demand, revenue, outcomes, or behavior. I document assumptions, test against holdout data, and explain where the model is reliable and where it is not. You get a model and the reasoning behind it.
A rigorous answer to a specific question: does this A/B test result hold, are these groups meaningfully different, is this result real or noise. I select the appropriate test, run it correctly, and tell you what it means for your decision — not just whether p < 0.05.
A visualization layer built for the people who will actually use it. Oriented around the decision, not the data. Clear hierarchy, no clutter. Designed to be updated and referenced — not archived after one presentation.
Structured insight from unstructured text. Customer feedback, survey responses, support tickets — classified, clustered, or scored. I build the model, validate it, and give you output your team can filter and use immediately.
Sneha's forecasting model changed how we planned our production cycles. The dashboard she designed is still in daily use months later — it became part of our standard operating procedure.
I work with a small number of clients each quarter so the work does not become generic. If you have a data problem with a real decision attached to it — send me a brief description and I will respond within 48 hours.
Every serious inquiry gets a response.
Tell me three things:
Boston, MA · Remote engagements globally
Currently accepting Q2/Q3 2026 projects.