jegeronimo
File
Edit
View
Run
Kernel
Tabs
Settings
Help
Python 3 (ipykernel)
Code
about.py
experience.py
contact.py
In [1]:

About

In [2]:
James Geronimo

๐Ÿ‘‹ Hello! I'm James.

In [3]:

๐Ÿ‘‹ I am an undergraduate student at UC Berkeley pursuing Bachelorโ€™s degrees in Computer Science and Data Science.

๐Ÿ’ป My technical interests lie in software engineering, generative AI, and CS education tools. I am passionate about leveraging technology across healthcare, biotechnology, and education.

๐ŸŽ™๏ธ Would love to chat! See the Contact tab for my email and social media profiles.

In [4]:

Use โ†‘โ†“โ†โ†’ or WASD to move. Eat the food, avoid the walls and yourself! You can also move cursor over the topography to warp the terrain.

Click or press Space to start

Score: 0
In [1]:

Experience

In [2]:

Oracle

Sep 2025 โ€” Present

Software Engineer Consultant

Contract

Engineering an AI tutor built on Oracle Cloud Infrastructure to promote active learning for 60+ data science students.

  • Build an AI tutor web app on Oracle Cloud with an interactive chatbot that generates MCQs to enhance learning
  • Design a RAG model on Oracle Database using JSONโ€“relational duality, vector data types, and hybrid search
  • Engineer data pipelines, leaderboards, and dashboards to track user interactions and analyze performance
In [3]:

UC Berkeley Data 100

Jun 2024 โ€” Present

Lead Teaching Assistant

Part-Time

Teaching 1200+ students in the UC Berkeley's upper-division data science course, Data 100: Principles and Techniques of Data Science.

  • Teach 1200+ students in the upper-division course Data 100, covering data processing, statistical/probabilistic foundations, exploratory data analysis/visualization, feature engineering, dimensionality reduction, predictive modeling, and optimization
  • Streamline course website navigation by developing in Jekyll and automating deployment workflows through GitHub Pages
  • Lead discussion sections to 40+ students, host office hours, debug DataHub issues, and maintain general course infrastructure
In [4]:

Amgen

Jun 2025 โ€” Aug 2025

Data Science Intern

Internship

Created a text-to-SQL application to democratize access to marketing analytics for Amgen's Commercial Data & Analytics department.

  • Achieve 87%+ accuracy in text-to-SQL operations by building generative AI and testing frameworks in Databricks
  • Optimize Databricks SQL connection and executions by 86%+ in the worst case, reducing operations to 0โ€“80 seconds
  • Deliver a Streamlit application optimized with caching to save 500+ hours and automate 60,000+ queries every year
  • Leverage large language models (LLMs) to automate campaign analytics across 27 brands using distributed AI agents
In [5]:

AmigoAI

Dec 2024 โ€” May 2025

Software Engineer Intern

Internship

Led independent projects in computer vision/optical character recognition (OCR), webscraping, and automation testing.

  • Devise and build an automated end-to-end testing framework using Playwright to detect regressions and improve test coverage
  • Engineer a robust and dynamic web scraper that mimics real-user behavior by emulating user agents, preloading consent cookies, and bypassing JavaScript-rendered content through direct HTML data extraction using Scrapy, BeautifulSoup, and Selenium
  • Scrape 1000+ data entries into JSON files for fine-tuning a custom-built LLM for standardized exam question generation
  • Construct a cross-format file processing workflow to extract text from files using PyMuPDF/Fitz and Pix2Text
  • Architect a computer vision pipeline that tracks real-time user progress to contextualize query processes in a fine-tuned LLM
In [6]:

UC Berkeley Data Science Modules

Jan 2025 โ€” May 2025

Software Developer

Part-Time

Built educational data science modules through UC Berkeley's Data Science Modules Team for El Camino College courses of varying domains.

  • Partner with four El Camino College professors to create educational data science modules teaching probability distributions, data visualization, classification, statistical modeling, and business analytics under the Data Science Modules team
  • Introduce 500+ students every semester to fundamental programming practices, statistics, and data-driven decision-making
In [7]:

UCSF Abbasi Lab

Feb 2025 โ€” Apr 2025

Research Assistant

Apprenticeship

Evaluated large language model (LLM) robustness via conversational red-teaming and multi-turn jailbreaking techniques.

  • Evaluate large language models (LLMs) robustness via red-teaming and multi-turn jailbreaking techniques under Abbasi Lab
  • Build an agentic LangChain framework to automate multi-turn LLM conversations and LLM-as-a-judge evaluations with GPT-4o
  • Customize and apply 10+ prompting goals and target strings from the MedSafetyBench dataset for attacker and target LLMs
In [8]:

Universal Health Services

May 2024 โ€” Aug 2024

Data Science Consultant

Contract

Constructed AutoRegressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) models for financial forecasting.

  • Achieved <18.5% and <25% average error in financial forecasting by deploying AutoRegressive Integrated Moving Average (ARIMA) and Long Short-Term Memory (LSTM) neural network models, optimized with extensive EDA and fine-tuning
  • Delivered Pyramid Analytics integration, Python script, and walkthrough video; leveraged mentorship from C-Suite executives
In [9]:

UC Berkeley L&S 22

Feb 2024 โ€” Aug 2024

Software Developer

Part-Time

Developed data science modules on Simpson's Paradox for UC Berkeley's L&S 22: Sense and Sensibility and Science.

  • Design and construct two data science modules deployed through Jupyter for the undergraduate interdisciplinary course LS 22
  • Introduce 500+ students every semester to fundamental programming practices, statistics, and data-driven decision-making
In [10]:

UW School of Pharmacy

Jan 2024 โ€” Apr 2024

Research Assistant

Apprenticeship

Leveraged Natural Language Processing (NLP) to carry out topic modeling and sentiment analysis on textual data.

  • Built and optimized a latent Dirichlet allocation model to cluster textual data, resulting in a coherence score of ~0.684
  • Leveraged natural language processing tools to enable complex analysis of textual data beyond human observation
In [1]:

Contact

In [2]:
primary_contacts = { "calendly": "calendly.com/jegeronimo/chat", "email": "jegeronimo@berkeley.edu" }
display(primary_contacts)
Out[2]:

Calendly

Let's Chat

Email

jegeronimo@berkeley.edu

In [3]:
social_profiles = { "linkedin": "@james-geronimo", "github": "@jegeronimo" }
display(social_profiles)
Out[3]:

LinkedIn

@james-geronimo

GitHub

@jegeronimo

In [4]:
resume = { "name": "James Geronimo", "format": "PDF", "status": "Available" }
display(resume)
Out[4]:

Resume

James Geronimo

jegeronimo Simple 1 3 Python
Ln 1, Col 1 Spaces: 4 Ready