Mountain View, California, United States
Interested in software engineering and AI
● Fine-tuned Meta’s Llama-3 on a synthetic text-to-SQL dataset utilizing the Unsloth library, achieving 65% accuracy on the test set, leading to significant performance gains in automated SQL generation. ● Developed a Streamlit-based automated data analysis and visualization tool using Mistral LLM, reducing manual effort. ● Led a team of six to design and implement a three-step automated data pipeline for cleaning, analysis, and ML modeling.
● Engineered and deployed a Dash-based web application for dynamic visualization of financial metrics, achieving a 50% improvement in error detection and resolution timeframes, and significantly enhancing data interoperability. ● Developed a production-grade, object-oriented data ETL pipeline in Python that consolidates and transforms data from multiple sources in seconds, boosting efficiency in data visualization and analysis processes. ● Authored extensive and accessible documentation for both the web application and data ETL pipeline using MkDocs.
● Designed an advanced Excel model for a Fortune 500 biotech firm's product to precisely estimate the Average Selling Price (ASP), achieving an error rate of just 0.01%. ● Constructed a tailored Excel workbook to identify the most cost-effective insurance plans, saving thousands of dollars per patient. ● Formulated a strategic action plan to mitigate risks from a competitor’s launch, leveraging detailed Excel analytics to devise optimal contracting strategies, securing multimillion-dollar savings.
● Refined an Excel model to strategically identify potential contracting entities under specific scenarios, optimizing for coverage and cost-efficiencies and yielding multimillion-dollar savings. ● Designed and implemented an Excel-based contract tracker empowering users to efficiently modify contracts, track existing brand-level contracts, and assess brand coverage across diverse geographical regions within the USA.