Andy Mahdavi Email & Phone Number

San Francisco, CA, US

San Francisco, CA, US

We're the upstart insurtech that's making real estate transactions close faster and more efficiently using machine learning and LLMs. I work in service of the software engineering, data science, data engineering, technology operations, cloud infrastructure, and technical project management teams at Doma.

San Francisco, CA, US

• Doma rose from startup in 2016 to #8 nationally in 2018, went public in 2021, and made it to Inc Magazine's list of top places to work the same year. Homeowners across America are closing their mortgages faster and.
• Are the first to deploy a property-based risk model in this industry (we've patented it!)
• Are the first to deploy transformer-based deep learning networks to automatically identify, process, action, and respond to real estate customer communications and hundreds of document types
• Enjoy complete ownership of model deployment, from scientific invention to business-relevant thresholding to the actual API call that touches our product
• Meet with deep legal minds in academia and the title industry to quantify risk and produce accurate loss forecasts
• Make virtuosic visualizations, from sliced 4-dimensional heat maps for our team's use, to intuitive business distillations of model lift for stakeholders

Guiding Auditive's AI/ML and overall company strategy as they bring third-party risk assessment into the modern age.

Mclean, VA, US

• I was the lead scientist within a group of vibrant product managers, engineers, and data scientists working together in cross-functional squads to enable machine learning on our customer acquisitions platform. We.
• Designing scientifically driven experimentation and engineering patterns for boosting machine learning power
• Creating an active ML framework that ties together business, governance, and systems monitoring
• Inspiring an open- and inner-sourcing mindset and helping to drive judicious combined use of tools such as Apache Spark, H2O, and scikit through pointed and technically deep consultations across a variety of languages.
• Theoretical research into pruning and traversing of high dimensional feature spaces
• Mentoring the growth of rising data scientists from theoretical, software engineering, as well as personal development points of view

Mclean, VA, US

• I served as a lead data scientist for several key model builds in the fraud detection and credit decisioning space, leveraging one of the largest training data sets ever used at Capital One and the first massively.
• Relentless pursuit of model lift by deep and joyful collaboration with an outstanding team of data scientists across feature engineering, pipeline architecture, and statistics
• Vetting of tech and data science startups in an advising role for Capital One Growth Ventures, our VC arm
• Creation of data science pipelines involving Hadoop MapReduce, H2O, Spark, Python, Scikit, vowpal wabbit, and Word2Vec, and of model serving systems involving Apache Storm
• Assembly of a thoroughly cleansed, public-facing transaction data base for use in live data science competitions, graduate student statistics face-offs, and physical data art that will soon grace our HQ lobby
• Shaping of intellectual property discussions in Capital One Labs

San Francisco, CA, US

• I was a tenured professor with time equally split between research (implementing parallel algorithms to retrieve dark matter properties using big data from space telescopes) and teaching (lecturing on physics and.
• Released open source OpenMP+MPI parallel implementation of Bayesian Markov Chain Monte Carlo code
• Led the development of proprietary GPU-accelerated galaxy cluster plasma integration code on the SFSTAR computational facility
• Mentored award-winning graduate students and placed them into PhD programs and jobs in Silicon Valley
• 7+ peer-reviewed journal publications as Andisheh Mahdavi

San Francisco, CA, US

• I was a tenure-track professor with time equally split between research (highlighted on NPR; see link below) as well as teaching. In addition, I led the creation of a new high performance computing facility and.
• Conducted high precision dark matter mass measurements through a multi-disciplinary data pipeline that combined X-ray, optical and radio wavelength telescopic data
• Built and led the direction of the on-prem SFSTAR high performance computing facility
• Created a new undergraduate course on the Big Bang for physics majors, focusing on the history of cosmic origins, from ancient origin myths to a calculus-based treatment of the historical unfolding of cosmology
• 14 peer-reviewed journal publications as Andisheh Mahdavi

Victoria, BC, CA

• Created distributed Levenberg-Marquardt minimization code using error-corrected numerical derivatives
• Used the XMM-Newton telescope to trace faint plasma in galaxy group environments
• 8 peer-reviewed publications as Andisheh Mahdavi

Honolulu, HI, US

• Used the world's largest telescopes on Mauna Kea, Hawaii to trace the velocities of galaxies in groups and clusters, leading to dark matter mass measurements
• Built deprojection tool for modeling galaxy cluster atmospheres in three dimensions
• 8 peer-reviewed publications as Andisheh Mahdavi

Garching, DE

• Developed sophisticated spatial feature and event detection technique customized for extremely low signal-to-noise data
• Used orbiting ROSAT X-ray Satellite to trace the link between cosmic plasma and dark matter
• Peer-reviewed publication

Phd, Astronomy And Astrophysics

Ma, Astronomy And Astrophysics

Ba, Astronomy And Astrophysics