About Us
Rho is the modern banking platform built for startups. Open accounts in minutes, issue cards, manage expenses, pay bills, and close the books – all in one connected platform backed by real human support.
About the role
You’ll work on high-impact data projects to help Rho detect, prevent, and better understand customer behavior—ranging from identifying early churn signals to mapping growth opportunities. This role involves designing experiments, building predictive models, extracting insights from unstructured data, and working with large, complex datasets. You’ll collaborate across teams to drive workflow efficiency, improve customer retention, and influence product direction, all while taking full ownership of your analyses and communicating your findings clearly to both technical and non-technical audiences.
General Overview of Potential Projects
New churn leading indicators. The current set catches a lot but not everything. Hunt for earlier and cleaner predictors of account death and treasury drawdown. Test them, backtest against known outcomes, graduate what holds.
New expansion and deposit-growth signals. The upside side of the book is less built out than the churn side. What predicts an account is about to move more money onto Rho, hire, raise, or expand its treasury? Generate and test candidates.
Unstructured data as a new signal source. The entire current signal universe is the warehouse. Transcripts are still an area we haven’t touched. Build LLM extraction experiments to pull signals that will never appear in transaction data: competitor mentioned on a call, product-limit frustration, expansion intent voiced directly. Different modality, genuinely additive.
Probabilistic modeling. Signal-interaction modeling (feature importance over stacked signals with SHAP) to replace hand-tuned tier overrides with measured weights. Behavioral clustering to find account archetypes.
Graph and network signals. Shared-investor and vendor co-occurrence structure for fundraise-contagion detection and referral clusters. Untouched today.
Iterations on the Markov Chain for customer health
Precise analysis on how AMS and GAEs are responding to clients in regards to churn signals and what playbooks could be improved.
Increasing GTM workflow efficiency
Analyzing production adoption, finding patterns that can increase it based on previous activity or detect likelihood to adopt → to make customers stickier.
You have:
Challenging coursework in Computer Science, Mathematics, Statistics, Data Science, or a related quantitative field
Project experience in statistics, ML, econometrics, or a related quantitative field
Proficient in Python, Comfortable with SQL
Willing to run a high volume of experiments and work with messy, incomplete data
Communicates quantitative work clearly to technical and non-technical stakeholders
Strong fit if:
Comfortable with ambiguity. Problems arrive underspecified.
Take ownership. You run your experiments end to end and do not need handholding. If something is broken or unclear, you chase it down rather than wait.
Question everything. You do not take a number, a signal, or an assumption at face value, including your own. You pressure-test before you trust.
Very fast learner. You pick up new tools, new data, and new methods quickly and independently, and you are not thrown by unfamiliar territory.
High attention to detail.
Deeply analytical. You reason from the data, quantify your claims, and can explain why something works or does not.
High throughput. You would rather run five experiments this week than one perfect one next month.
Hourly: $20-$35 | Start: ASAP