Soren Rajani

About Soren Rajani

My professional and academic experience include:+ LLM fine-tuning and Prompt Engineering+ e2e ML forecasting and model deployment+ Causal Inference and statistical modelling+ Data Extraction via Multi-modal AI+ Design of ExperimentsCurrently working as a data scientist specializing in time-series forecasting and design of experiments. I work with docker and airflow to deploy python models at cadence. I also build efficient data pipelines integrating dbt and snowflake via sql for efficient data ingestion and feature engineering.My past research involves computational efficiency for LLM fine-tuning utilizing gradual unfreezing and linear probing approaches. I am also involved in an experimental data-scraping technique for adversarial settings using multi-modal AI. If you have an interest in either of these fields please reach out to connect.I have experience working in economic research and have completed projects with observational data for causal interpretation. I conducted my undergraduate thesis on the impact of economic openness on the link between pay and productivity using Python and LaTeX. I have extensive experience with python and R using libraries for machine learning.

Experience

1. Data Scientist

   BetterHelp

   Jul 2024 — Present · Mountain View, CA, US

   Reducing time to decision by implementing cutting-edge experimental design strategies+ Conducting data analysis in the mental health therapy industry to derive actionable insights+ Deploying and designing multiple time-series forecasting models+ Designing and executing A/B tests to provide data-driven recommendations+ Using short term signals to predict long term experimental outcomes with surrogate modelling

Education

• The University of British Columbia

  Bachelor of International Economics and Minor in Statistics

• University of California, Berkeley

  Master's, Statistics