Nice to meet you.

I'm Brandon.

Am I keeping up with AI… or is AI catching up to me?

Every week, I feel like the world was built just a little bit more for people like me. People with creative ideas and technical curiosity, active explorers with a sense of beauty and a will to win.

What it takes

It has never been easier to get a prototype up and running, but it has never been more challenging to build a viable product that won't be stolen from you.

In that kind of environment, you want someone who is completely immersed in the tech, who has a working theory built on personal experience and facts. That person will have an intuition for what's really possible now — and grounded technical reasoning for what will likely be possible a year from now. Only a person like that will be able to prioritize the features and understand the largest risks to your product.

Which are:
01 Obsolescence before UAT
02 Second mover (first loser)
03 Your plans were made on hype
04 …and everything else your industry has worried about for decades

Thank you for taking the time to read. Please enjoy the experience of being here, and connect if you see value.

I grew up competing. Nationally ranked fencer through high school and college — the kind of sport where you read your opponent in real time, make decisions in fractions of a second, and learn that preparation and instinct aren't opposites. Princeton sharpened that into something else: a Comparative Literature degree that taught me how meaning is made, how language carries weight, and how to think across disciplines without losing rigor.

Then I spent a decade writing. B2B copy, brand voice, content strategy — for tech companies, telecoms, and a boutique agency where I eventually became the person they called when something was broken and no one knew why. I got curious about the machinery behind the words. Curious enough to pull myself into it.

Technology drew me in because it has both sides — the creative and the technical — and the best products live at that intersection.

At Wharton, I've owned AI products from concept through production: a custom course chatbot built on a vector DB of class transcripts, a private elections platform scaled through two successful releases, a school-wide attendance system that shipped a feature the vendor now carries in their production app. I sit with engineers, InfoSec, faculty, and executives — and I speak all of those languages well enough to actually move things forward.

I'm second author on a published paper about the Lecture Recall project. I co-chair the AI Special Interest Group and the AI Community of Practice at UPenn. I've given talks on GenAI to audiences from MBA students to IT staff conventions to DataPhilly. I am not a spectator in this moment — I am actively building in it.

What I'm looking for next is an AI product that deserves the kind of ownership I know how to give it. Fast-paced, competitive, consequential. I know what that feels like. I'm ready for it again.

Things I own — built, shipped, and still responsible for.

Lecture Recall
Wharton School  ·  2023 – Present
Wharton's custom AI course chatbot — built on a vector database of class transcripts so students can retrieve and query lecture content in natural language. I ideated the prototype as a Custom GPT with actions to a vector DB, led it through beta testing, and am now building v1.0 for Fall 2026 delivery. Coordinated with the AWS Innovation Center on a hybrid architecture that was the first of its kind in higher education. Worked with InfoSec throughout to navigate approval in a complex institutional environment. I'm second author on the published paper.
AWS Bedrock RAG / Vector DB Prompt engineering Python
Published research First hybrid RAG in higher ed v1.0 → Fall 2026
Facelect
Wharton School  ·  2022 – Present
A custom, private elections application for Wharton — maintained and expanded through two successful releases with usage doubling from 2024 to 2025. Delivered on a $3k total development budget with minimal use of in-house dev resources. I own the roadmap, manage the release cycle, and drove the onboarding work that produced the usage growth.
Product ownership Release management Vendor coordination
2× usage growth 2024→2025 2 successful releases $3k budget
Attendance & Class Recording
Wharton School  ·  2022 – 2024
Led school-wide vendor onboarding to replace an in-house attendance product — from inception through successful deployment across Wharton. Identified a feature gap and designed a custom absence-request workflow that the vendor now ships in their production application. Simultaneously salvaged a parallel in-house build into a well-reviewed Class Recording Permissions product.
Vendor management Product design Stakeholder alignment
School-wide deployment Custom feature adopted by vendor Two products launched

Strategic initiatives, partnerships, and complex org problems I've navigated.

Wharton / UPenn
Ethan Mollick GenAI Lab
2024 – Present

Strategic partner to Ethan Mollick's Generative AI research lab — consulting on infrastructure for a unique multi-turn prompt testing application. Liaised between research, engineering, and institutional stakeholders. Also served as lecturer for an infrastructure session of The Studio, a masters-level program.

GenAI infrastructure Research partnership Guest lecturing
Ongoing strategic partnership Infrastructure lecture · The Studio Spring 2026
Wharton
ADIEU — Active Directory Migration
2023 – 2024

Built and maintain automated data pipelines from the Calendly API and Wharton information systems into Excel — helping coordinate front-of-house staff as they plan and execute a complex migration of Wharton-domain machines to Kite. Reduced manual coordination overhead and gave operations teams visibility they didn't have before.

Python automation ETL pipeline design API integration
Reduced manual coordination overhead Ops visibility unlocked
STRATACACHE
Digital Marketing & Infrastructure
2021 – 2022

Brought in as a content marketer, I expanded into technical ownership across multiple fronts: created a content taxonomy for the CMS, led the consolidation of two brands' DNS under a single provider, and diagnosed a broken custom Google Tag Manager instance without documentation or access to the original developers. Wrote and maintained a Python integration script for automated content publishing on Contentful.

Python scripting CMS architecture GTM / analytics DNS management SEO / PPC
Conversions 58→85–94 per quarter Diagnosed broken GTM — no docs, no original devs
Oct 2025
Exploring Wharton's Lecture Recall Tool: A Conversation with Pete Fader
Instructional Design Seminar
May 2025
Wharton GenAI Update
UPenn IT Staff Convention
Apr 2025
Workshop: Local LLMs
Wharton Computing / WRDS
workshop
Jul 2024
Slow Down Your Chatbot Build
UPenn AI Special Interest Group
Jun 2024
Failing 10× at Machine Learning
DataPhilly
2023–24
What Is Generative AI?
MBA Office of Career Management · Wharton HR · WRDS · UPenn Alumni Relations

A lab notebook. Essays on philosophy of technology, experiments with AI tooling, half-formed products, published research — whatever I'm turning over. Some of it is finished. Most of it isn't. All of it is honest.

Essay
What does it mean to think with a machine?
Philosophy of technology meets the LLM moment. Where does extended cognition end and outsourcing begin — and does the distinction matter?
in-progress
Experiment
Claude vs. Codex vs. Opencode: a practitioner's notes
Not a benchmark. A working log of how these tools actually behave when you're trying to ship something real — what they're good at, where they break, and what that tells you about the underlying architecture.
// ongoing
Product idea
The case for async AI office hours
Universities spend enormous resources on synchronous advising that scales poorly. What would a product look like that used RAG + persona modeling to make expert judgment available at scale — without pretending it's the real thing?
in-progress
Research
When to slow down your chatbot build
The talk that became a framework. Institutions rush to deploy AI and skip the hard questions about accuracy, trust, and maintenance. Here's what Lecture Recall taught me about building AI products that actually hold up.
writing-up
Experiment
AWS Bedrock for the time-constrained PM
What you can actually build with Bedrock if you're not a full-time ML engineer — and what still requires one. Notes from building Lecture Recall without a dedicated data science team.
drafting
Essay
Technology and the shape of attention
Something in the philosophy of technology lineage — Heidegger, Stiegler, Ellul — feels more urgent now than when it was written. An attempt to bring those frameworks into contact with the current AI moment.
outline
// more arriving  ·  this is a live space, not an archive

I'm looking for an AI product role where ownership is real — not a title attached to a committee. Fast-moving, technically grounded, consequential.

Also happy to talk: philosophy of technology, the state of AI tooling, what it actually takes to ship an AI product in a complex org, or fencing.

Start a conversation