- Created Athena, a web application with two main features: the Report Generator and "Chat with PDF" from scratch in 8 weeks, being the sole developer and functioning as the project manager- The Report Generator creates an AI-generated report on any company when supplied the name and domain. The sections you can create include: Company Description, Product Overview, Market Overview, Key Competitors, Jarvis Info (from internal CRM), Investment Theses, Key Risks, and Questions for the CEO- Chat with PDF accepts TXT, PDF, PPT, XLSX, CSV and more file types as input, and allows an LLM to use chunks of the document as sources when answering queries to prevent hallucinations. This feature incorporated "Short-Term" and "Long-Term" memory, which utilized a transient, in-memory vector database, and a VDB that contained all historically uploaded documents- Athena's architecture is a frontend React application hosted on Vercel, and a backend FastAPI API hosted on Azure, both including production/development environments, CI/CD pipelines, and Auth0 authentication/authorization- Wrote API, frontend, infrastructure documentation and handled the handoff of the application to a 3rd party tech team. Also created GitHub organization, repositories, Azure/Vercel/OpenAI/Pinecone accounts- Wrote two MVPs in Streamlit and Shiny for Python, before refactoring to React + Python in last 4 weeks

- Developed full stack web application for energy team to qualify sites, prospect for client contact information, and model energy consumption for newly-constructed buildings. - Integrated internal data with external data from API calls and web scrapes to create end-to-end site qualification pipeline and contact discovery tool in user-friendly frontend. Reduced manual work required by sales reps for prospecting from an hour-long flow to a 2-minute process- Re-modeled project pricing tool using VBA Excel to automate pricing of 7 distinct project types and reduce human error, reducing a 2 hour process to 5 minutes, and 7 separate tabs into one main dynamic tab.- Performed internal operating margin analysis and expanded model to show margin breakdowns for solar and storage hardware, and EPC margins

Analyzed internal order and seeding data in addition to external weather, population, and market data to remodel our current forecasting techniques to more accurately predict the quantity of product requested by our largest customer, ultimately improving model accuracy by 10% resulting in 500 lbs of product saved per week. Developed a data ingestion, cleaning, and modeling pipeline in Python to employ statistical forecasting and machine learning methods to reduce excess seeding of product and identify factors most significant to the fluctuation of orders received from our retailers. Working cross-functionally with sales, marketing, and upper management teams to understand entire life cycle of our produce and understand how data can be best used to help put high-quality microgreens on peoples' plates.

Developed custom software to scrape the internet and parse incoming data to augment database with over 600,000 entries. Designed web crawler to retrieve thousands of company links with nested searches using requests library in Python.

Developed web crawler to automatically scrape international oil production data from 117 countries from 2004 to present. Designed financial models to forecast competitors’ future shipping capacity and expected earnings.

Automated the preprocessing and categorization of thousands of entries from clients with neural network using keras and word2vec libraries in Python. Analyzed historical client data to visualize growth of targeted advertisements provided to over 80 million homes.

Taught extracurricular courses in STEM for K-8th students focused on math, LEGO robotics, and computer programming.