Technical Content

Training Deck Builder – Automation Tool, Python/PowerPoint

I identified a manual bottleneck in our role-based training development process. Our team was hand-assembling PowerPoint decks for 28+ roles by cross-referencing an Excel spreadshhet against dozens of individual chapter files stored across multiple SharePoint libraries. It was a tedious, error-prone process that scaled poorly. I scoped the requirements, designed the workflow, and collaborated with Claude Code to build a desktop tool that automates the entire process. I drove the design decisions such as authentication approach, matching logic, file architecture, and how to handle missing content while using the AI to accelerate the Python implementation.

Highlights
- Reads a Excel file to determine which chapters apply to a given role based on responsibility assignments
- Searches locally synced SharePoint folders for matching chapter PowerPoint files using fuzzy name-matching
- Compiles matched chapters into a single role-specific training deck using PowerPoint COM automation (preserving all formatting, images, and layouts)
- Outputs the compiled deck to a synced SharePoint folder for automatic team access

Impact: Reduced what was a 30–60-minute manual assembly task per role to a single button click across 28 roles, while ensuring nothing gets missed. The tool is designed to outlast my tenure on the project.

Process Finder User Interface, Power Automate:

The Process Finder is an interactive, searchable index that surfaces clinical operations process maps, role-specific workflows, and linked training materials in one place. It replaces the need to dig through disconnected SharePoint folders, PDFs, and spreadsheets to find what to do, who owns it, and where the training lives. Built as an HTML prototype to validate the design before a Power Apps deployment, it demonstrates how structured process data can become a self-service tool for clinical team members.

Highlights
- Filters processes by role (20 clinical ops roles like CTC, CTL, QOC, Business Admin) so users only see workflows relevant to them
- Provides searchable gallery cards organized by study lifecycle phase (Start-Up, Maintenance, Closeout)
- Renders interactive branching workflows where users click through decision points (Create or Update? Which type?) to follow only their specific path
- Links directly to training QuickCards and resources on SharePoint, opening in-browser via Word Online

Leverage LLM Tools to Enhance Content Quality and Productivity, Claude Code:

Where I see a need, I use LLMs to build tools and enhance processes that help ensure I am doing a full doc review.

Highlights
- Built a Role Glossary Lookup Tool using Claude Code that allows you to verify role names and definitions while reviewing SOPs and WIs without leaving Microsoft Word.
- Created an Undefined Abbreviation Detection Macro using Claude Code that identifies and highlights abbreviations that have not been defined in a document, improving review efficiency and document quality.

Impact: Collapsed 5-6 open tabs into a single, streamlined user interface that surfaces the right content to the right user at the right time.

Continuous Learning of Document Management Systems:

Moving into the medical field from software meant learning a whole new set of document management systems that are standard in the industry. I've embraced this challenge and continue to dig into the nuances of each, enabling me to better support my stakeholders.

Highlights
- Massive increase in my knowledge of Veeva Vaults, including VeevaQuality, VeevaClinical, and VeevaRIM One
- Increased use and understanding of SharePoint functionality
- Learning Microsoft Power Apps

Training Development & Knowledge Enablement:

I enjoy creating training resources that help people understand complex processes and work more independently. By combining technical writing, instructional design, and process knowledge, I develop learning materials that improve user understanding, support consistent execution of workflows, and reduce the need for one-on-one support.

Highlights
- Develop and maintain training content in Brainshark and Rise 360
- Create training videos in Clipchamp to guide document owners through Veeva Workflow Quick Cards and document workflows
- Design user-focused training materials and guidance that improve knowledge transfer and workflow adoption

Docs-as-Code Knowledge Base: Campo de Estrellas Knowledge Base

I built and deployed a docs-as-code knowledge base using Docusaurus (React) and GitLab CI/CD, configuring Node 20 build environments, YAML pipelines, and GitLab Pages for automated production releases.

This is an internal knowledge base of SOPs accessible to collaborators for improved knowledge sharing.

For the knowledge base homepage, I featured our most-used procedures as well as frequently asked questions by our families for easy access. Each procedure loosely follows the Every Page is Page One guidelines of topic-based writing and linking richly across content so that forms, affiliated procedures, and guidelines are quickly accessible.

Highlights
- Structured content architecture
- Sidebar configuration and content organization
- Version-controlled documentation
- CI-based static site deployment

API Documentation I created

- Webhooks Schema
- Dangerous Goods
- Create a Label

User feedback dashboard and implementation system:

This project revolutionized how the Product Content team handled user feedback by transforming a cumbersome manual process into an automated, efficient workflow. By integrating Airtable, Jira, and Zapier, I created a system that not only saved time but also provided deep insights into user sentiment and article performance.

Highlights
- Within one quarter of launching, we saw a drop in negative feedback from 451 responses to 111 responses.
- Additionally, we were able to see metrics for positive feedback, a score we had no way of measuring previously. The positive feedback averaged between 60-75% for the year 2023.

Training Chat Agents:

I trained several chat agents on ShippingEasy documentation, learning early on the way LLMs consume, search, and surface content. I prepared a series of FAQ articles that we fed to the agent. There was also a Slack plugin that support could use to ask it questions, which I would monitor for the effectiveness of the responses. These were early versions of the built-in chat agents we see everywhere today, and this was my first experience honing content to be recognized by machines.

Highlights
- Ultimate AI 2023-2024
- Sunshine 2020-2023
- Talla 2018