Jake Brophy, M.S. Email & Phone Number

Los Angeles, Ca, Us

Davis, California, United States

• Led the development of an AI-driven medical search engine at a startup, architecting and coding the entire backend in Python, and deploying it via a Flask app on Azure web apps.• Designed a sophisticated agent system using multiple language models to automate complex tasks such as medical knowledge retrieval, case study comparison, and drug interaction analysis, significantly enhancing the medical accuracy and reliability of the language model system.• Integrated API calls, web scraping, and a vector database to efficiently gather, parse, and summarize extensive medical data into actionable insights, improving the accuracy and usability of the AI system’s insights.• Accelerated the development cycle, launching a fully-fledged product ahead of schedule. See the live application at clarifi-tech.com

San Francisco, California, United States

-Fine-tuned Open-source and private LLM's using QLoRA on tasks like visual story content conversion-Provided monthly optimal budget allocations to marketing using MMM methods-Deployed fine-tuned language models using HuggingFace, AWS Amplify, and SageMaker

San Francisco Bay Area

Building and Deploying Time Series Forecasting Model for a mobile gaming company.• Built CNN-LSTM for 7-day organic downloads forecasting, as well as XGBoost for aggregate weekly revenue forecasting. Achieved sub-10% MAPE, and deployed models using Jenkins and AWS.• Used Facebook Robyn MMM in R to generate optimal budget allocation across 7 different ad channels, optimal budget is forecasted to increase return on ad spend by 5% over current allocation.

As part of the MSBA, building and optimizing data pipeline for property management startup.• Led a team of student data analysts and engineers to accomplish a diverse set of goals that ranged from building data pipelines to conducting financial analysis.• Building real-time data pipeline using AWS Kinesis to eliminate dependence on third-party ETL tools using Python, SQL, AWS, and Google Cloud. Final product will increase pipeline scalability and reduce company expenses.• Built, tested, and optimized spatial econometrics model in R and Python to understand the financial impact of customer reviews on listing price.

San Francisco, California, United States

Supporting faculty in investigating the formation of echo chambers and spread of hate language online.• Used Huggingface Transformers library and PyTorch to fine-tune Falcon and LLaMa2, state of the art Large Language Models, in explainable hate speech detection. • Built neural network and other NLP models in Python for text classification using TensorFlow, achieved validation set accuracy of 92% using a CNN and Google’s transformer-based algorithm BERT.• Combined neural network language models with a propensity score matching causal inference framework for 50k+ users and 20 mil+ tweets to analyze the impact of an online community on the content an individual puts on the internet.

Los Angeles County, California, United States

Assisted a PhD candidate in writing publication on forecasting the cross section of Cryptocurrency returns.• Used Python to clean large and complex financial datasets enabling feed into neural networks. This helped streamline work process and bring focus to more complex parts of paper.• Helped develop and test benchmark deep learning forecast model.

Los Angeles County, California, United States

Used High School golfer stats to predict future college All-Americans in order to inform UCLA golf’s recruiting efforts. I also aided in the building of an R Shiny app to easily access this data. Built several statistical models such as CART and Logistic Regression, then evaluated each model’sperformance using a train test split. Webscraped golfing data to keep the model up to date

Los Angeles County, California, United States

Used Python to investigate fight data across time to see how the sport of MMA has changed, looked at factors such as fight stance and height as well as changes in pay per view purchases and compiled our findings in an article. Used data visualization tools like matplotlib and seaborn.

Los Angeles County, California, United States

Analyzed business opportunity data for marketing tech startup that generates affiliate revenue for publishers.• Mapped out the Business Development Team’s pipeline from initial outreach to close in order to inform which outreach methods were the most effective, presented findings to leadership. • Automated lead detection in Python, saving the team an average of 4 hours per day and generating 36000 leads. Also discovered new publisher partner that the team was able to sign and move closer to quarterly revenue goals.

Los Angeles, California, United States

Greater Los Angeles Area

I am an opinion writer for the Daily Bruin at UCLA. Responsibilities include networking with potential sources, staying up to date on campus events, and producing an opinion piece ever 2 weeks.

Greater Los Angeles Area

Master Of Science - Ms, Business Analytics

Bachelor Of Arts - Ba, Economics, 3.875/4.0