Golnoosh Golpayegani Email & Phone Number

San Francisco, CA, US

San Francisco, CA, US

Irvine, California, US

• Pioneered a collaborative filtering model to personalize the Marvel Snap shop, elevating Marvel Snap’s KPIsand securing the "Best Mobile Game" title at The Game Awards 2022 and the 2023 Apple Design Award.
• Implemented a hypothesis-driven impact assessment framework, empowering cross-functional teams to evaluate feature success and enhance workflow efficiency in a dynamic startup environment.
• Built a predictive LTV model for Marvel Snap players, segmented by regions, channels, and install cohorts, enabling the marketing team’s user acquisition budgeting and aligning decisions with investor goals.

Culver City, California, US

• Prototyped and deployed a new gradient boosting multi-classifier model to suggest personalized bundle offers to game users in order to optimize their experience and increase their overall revenue of Disney Emoji Blitz.
• Significantly improved Panda Pop gamer experience by productionizing skill-based, scalable clustering techniques and using dynamic time warping time series predictions that lifted the daily retention by 2.5%.
• Led feature engineering for predictive modeling, working closely with stakeholders and data engineers to ensure optimal performance. The engineering led to a 20% increase in LTV accuracy.
• Designed and feature-engineered a Spark-based gradient boosted trees payer propensity model to target Cookie Jam payer users and managed the end-to-end ML lifecycle using MLflow.
• Built data pipeline for a survival analysis model on Databricks platform to identify about-to-churn users of the Harry Potter: Hogwarts Mystery game and predict their age as an important use case for product decisions.

Berkeley, CA, US

• Built the state-of-the-art instrumentation and software with SETI (Search for Extraterrestrial Intelligence) in collaboration with scientists at the University of Oxford and UC Berkeley, and led the WVU team for.
• Implemented an autonomous prioritizer model to manage a large number of detected events by computing 400+ features during the figure generation to be used for building a random forest probabilistic classifier model.
• Discovered the first transient ever detected by ALFA receiver during the telescope slew among eight more FRB candidates, sampled the unexplored end of the FRB population, and calculated the expected FRB rate based on.

Berkeley, CA, US

• Developed a GPU accelerated pipeline for finding astrophysical transients called "fast radio bursts" (FRBs) from the backend of the world's largest fully steerable telescope to convert the raw data to the necessary.
• Detected over 2000 single pulses which helped to validate the pipeline, and carried out blind injection analysis of the data to assess the completeness of the pipeline and quantified the performance of the pipeline.
• Revised the current rates and the source count index of FRBs through a Monte Carlo Simulation at 95% confidence interval based on the null results and cross-checked them against 13 other relevant data sets, which.

Morgantown, West Virginia, US

• Led a team of 10 scientists around the globe to develop and maintain a real-time commensal operating FRB detector pipeline on the 20-m telescope at the Green Bank Observatory that processes more than 20+ TB data.
• Led the deployment of a unified python-based post-processing data acquisition framework for single-pulse search including the "friends-of-friends" clustering algorithm for discrimination of false positives and.
• Constrained an upper limit on the power-law index by modeling the FRB rate-fluence exponent at the 90% confidence level using the non-detection results.
• The meta-data from GBTrans backend was employed for training, validating, and testing a novel CNN algorithm, which has resulted in an open-source, state-of-the-art machine learning pipelines for FRB detection.

Morgantown, West Virginia, US

Taught workshops on the fundamental data and computational skills needed to conduct data-driven research.

Morgantown, West Virginia, US

• Developed the first validation framework for verifying and reporting FRBs in which FRB-like events can be evaluated as real or otherwise, by reanalyzing the examples of false-positive events in archival data, as well.
• Created a set of verification criteria to provide a robust statement to any potential FRB discovery by using the prototype theory of categorization. The results have been widely used by astronomers around the world.

Morgantown, West Virginia, US

• Taught upper undergraduate-level courses such as descriptive astronomy, modern physics, electromagnetism lab, and optics and provided regular feedback to 500+ students relative to their performance.
• Assisted more than seven instructors in developing new material and policies for courses to ensure content and delivery methods meet learning objectives.
• Mentored students to actively participate in all aspect of the experiments in the lab, including but not limited to, discussions, demonstrations, outside assignments, research, enrichment activities, etc.

Doctor Of Philosophy - Phd, Applied Physics

Mba, Information Technology

Master Of Science - Ms, Applied Physics