Rohit Kumar
Senior Data Scientist at Opendoor · 8 years · Python, SQL, PySpark
About
I am a Senior Data Scientist at Opendoor with over 8 years of experience in data science and machine learning, spanning domains such as Real Estate, Fintech, and Gaming. I have worked on impactful projects involving predictive and causal ML, MLOps, and LLM application development, delivering measurable business outcomes such as a 95% lift in contract conversion and ~$4M/year in profits. My strongest skills include Python, SQL, PySpark, MLOps, Databricks, Airflow, dbt, and Scikit-learn. I am currently looking to transition into a Data Engineer role, where I can leverage my technical expertise and hands-on experience to build scalable data infrastructure.
Experience
Senior Data Scientist
Sep 2024 – Sep 2026Opendoor
Owned Direct Mail targeting yearly roadmap — translating acquisition and GMV growth goals into priorities across ML infrastructure, monitoring, and scalability — scaling the underlying models into foundational ML infrastructure reused across multiple use cases org-wide. Built and deployed a suite of causal uplift and propensity models for Direct Mail targeting — including a mailer uplift model, Cash/Listing propensity variants, and True Seller propensity model (extended across digital channels like FB & Instagram), delivering 95% lift in contract conversion and ~$4M/year in profits. Architected and scaled production ML infrastructure — productionizing a combined uplift/propensity scoring pipeline on Databricks workflow with Delta Lake outputs and a dbt-based eligibility layer, scaling the model stack ~4x in data volume via distributed PySpark/XGBoost for error-free national expansion (~$500K/year). Redefined targeting pipeline's ranking objective from conversion probability to expected
Senior Data Scientist
Oct 2020 – Jun 2024Zupee, Gurugram
Implemented Media Mix Model (Robyn) to optimize CAC across acquisition channels, driving ›170K incremental FTDs/month within the first 2 months. Built a DWT-based framework to measure TV ad impact on acquisition, automated via AWS Glue/Tableau, reducing CAC by ~47% within a year. Built a Python framework and FastAPI dashboard to detect & reduce pre-experiment test/control bias from ~40% to ~5% with preemptive bias detection. Led Geographical Split Tests to optimize branding spend allocation. Applied causal inference (Propensity Score Matching) to quantify KPI lift (cricket day, holiday engagement drops, etc) increasing ~₹1.5Cr projected revenue. Integrated Google Ads Data Hub to build ADH-Python/AWS Glue BI pipeline on Tableau that surfaced insights on brand lift, creative fatigue, and branding/performance overlap to standardize measurement across branding levers. Led a 2-person Data Science team, owned quarterly/annual roadmaps, and served as SPOC for Data Science and Analytics across
Product Manager - Analytics
May 2020 – Sep 2020ICICI Bank
Identified segments based on interest profiles and created marketing strategies for auto loan products. Conducted AB tests on several web components like creatives, loan descriptions, banners, etc to identify a creative strategy to drive higher user conversion
Data Scientist
Jun 2018 – Mar 2020ePayLater
Built & deployed credit-risk models via Flask REST API on AWS Elastic Beanstalk, reducing delinquency from ~35% to ‹9% & increasing good-customer transaction volume by 30% in 6 months. Trained and implemented 4 predictive models to estimate recovery probability and collection effort, reducing monthly collection costs by ~45% in 3 months. Derived 20+ tables from unstructured data sources to streamline analytics and data science pipeline.
Education
Indian Institute of Technology, Roorkee
2014 – 2018Bachelor of Technology, Civil Engineering
Projects
Direct Mail Targeting Models
Python, PySpark, XGBoost, Databricks, dbt, Airflow
Built and deployed a suite of causal uplift and propensity models for Direct Mail targeting, delivering 95% lift in contract conversion and ~$4M/year in profits.
dbt/Airflow Feature Pipelines
dbt, Airflow, Snowflake, PySpark
Engineered dbt/Airflow feature pipelines from third-party Snowflake data, delivering ~$3M/year in cumulative profit.
Pre-Experiment Bias Detection Framework
Python, FastAPI, AWS Glue, Tableau
Built a Python framework and FastAPI dashboard to detect and reduce pre-experiment test/control bias from ~40% to ~5%.
Credit-Risk Models
Python, Flask, AWS Elastic Beanstalk
Built & deployed credit-risk models via Flask REST API on AWS Elastic Beanstalk, reducing delinquency from ~35% to ‹9%.
Skills
Certifications
AWS Certified Solutions Architect
2021AWS
Achievements
AWS Certified Solutions Architect
2021Certification for ‘AWS Certified Solutions Architect’
0 questions answered on readyforhires in the past year, 0 active days, longest streak 0 days.