Jump is looking to identify advisor best practices by researching hundreds of thousands of aggregated, anonymized advisor <> client conversations. The company needs a research intern to ensure their methodology for this research is as rigorous and reliable as possible.
Requirements
- Training in observational causal inference and causal machine learning.
- Strong foundation in statistical modeling and data analysis.
- Experience working with large datasets (Python, R, or similar).
- Familiarity with NLP or interest in applying LLMs to real-world research problems.
- Familiarity with behavioral science, financial services and causal ML libraries such as EconML, DoWhy, or CausalNex.
- Designing and applying robust causal inference strategies to observational data.
- Exploring causal ML approaches to uncover behavioral drivers of outcomes.
Responsibilities
- Apply observational causal inference methods with clear identification strategies to isolate conversational variables that causally influence outcomes.
- Engineer structured features from unstructured transcript data (e.g., advisor talk ratio, sentiment, interruptions, trust markers, hesitations) using LLMs, embeddings, and NLP.
- Analyze large-scale anonymized transcript datasets.
- Strengthen the methodological rigor of our research design and analysis.
- Contribute to research that pushes the financial advising industry forward.
- Develop a sustainable process and reusable causal model that the team can operate independently after the internship, ensuring continuity and scalability of insights.
- Building statistical and causal models to assess advisor effectiveness.
Other
- Graduate student (MA/PhD) or college senior in statistics. (Highly qualified juniors are also eligible, and we’re also open to those in applied math, computer science, or quantitative economics with applicable training.)
- Curiosity and exploratory creativity: the ability to go beyond validating predefined hypotheses and propose / uncover novel conversational levers.
- Intellectual curiosity and a passion for using data to drive impact.
- Commitment to methodological rigor and careful research design.
- Summarizing findings for both technical and non-technical audiences.
