Mission
For years, every time my team shipped something new, I would stop everyone and say the same thing: put it on your resume.
I said it so often it became a reflex. We work for our resume. We work for the marketability that our resume provides, which affords a continued and better standard of living for our families. The company can reorganize tomorrow, but what they cannot take away is what we built and what we know.
I meant it every time. And I said it for years. What I didn't realize was that I was describing a product I hadn't built yet.
Who I am
I'm Brian Lemus. I spent 21 years at Bank of America, most recently as a Senior Vice President leading enterprise data and AI. Over two decades I worked across six successive technology waves, from data warehousing and CRM analytics through regulatory detection, speech analytics, NLP and ML, and generative AI. Each wave built on the one before it.
My technical specialty is language. Specifically, using natural language to detect rare and sensitive events at enterprise scale: regulatory violations buried in millions of customer interactions, fraud signals in transaction data, compliance risks that surface only when you know what patterns to look for. That discipline, the obsession with finding what's true in a sea of noise, is what I brought to AI governance, and it's what I'm bringing to Career Vault.
The teams I led spanned over 150 people across data science, analytics, annotation, and technology operations in the US and India. Together, we deployed the company's first enterprise-approved generative AI model through 17 internal governance checkpoints and maintained zero regulatory findings across more than 15 federal examinations spanning a decade. We co-invented a patented machine learning system. I contributed to President Biden's AI Executive Order.
I share all of that not to impress you, but so you can verify it. My LinkedIn is here. The people I led wrote recommendations you can read. In a world where anyone can claim anything, I want you to be able to confirm who I am before you trust me with your data.
What happened when I tried to tell my own story
After 21 years, I left the bank to pursue AI research. And almost immediately, I ran into a problem I recognized.
My career was scattered across close to 100 documents that I had built up over hours and hours of conversation, trying to capture 21 years of experience without any real structure. No formal system. No template. Just me trying to get it all down. I had AI tools available, but the volume was so large that even the most advanced models would run out of their context window after a single pass through the material. Creating one resume took me 20 hours of feeding documents, cross-referencing facts, and fine-tuning language.
I got it down to 5 hours. But only after I solved the real problem: my data wasn't structured in a way I could trust. Facts were inconsistent across documents. Metrics conflicted with each other. AI tools, when I used them, would confidently present details about my own career that were wrong. A hallucination rate listed as 25% in one place, 5% in another, 0.05% in a third. The real number was under 0.5% material hallucination, meaning outputs completely void of actual events. I know because I built the taxonomy that distinguishes a helpful inference from a fabrication. Certifications I never earned appeared on drafts. Job titles I never held showed up. My own career was being hallucinated back to me.
I spent months building what I now call a career vault: a single structured source of truth, verified fact by fact, where every claim traces to evidence. Once I had that, everything changed. Every subsequent resume came from one trusted baseline. No more reconstructing from memory. No more hoping I remembered the right number. No more starting over.
That experience is Career Vault. I'm not building it from theory. I'm building it because I lived the problem, solved it for myself, and I know exactly how painful it is when your career data is scattered, unverified, and unreliable.
What I believe
Two books shaped how I lead. “Team of Teams” by General Stanley McChrystal taught me that shared consciousness and empowered execution, not command and control, are what make organizations adapt under pressure. The leader's job isn't to be the chess master controlling every move. It's to be the gardener creating the conditions for the team to succeed, and then asking one question: what do you need? Ryan Holiday's “The Obstacle Is the Way” taught me that obstacles are not interruptions to the work. They are the work. Every setback I encountered across 21 years, every regulatory challenge, every system that broke, became the thing that made the next system stronger.
I carried those principles through every role. When oversight groups needed to approve an AI model, they weren't treated as gates to get past. They were brought in at the beginning and made design partners. When the annotation team needed to label data for a new model, I sat down and did the labeling myself first, including labeling in Spanish, across three separate deployments. Not because a leader needs to do every job, but because I wanted to understand what the work actually felt like before I designed the process around it.
What my team says
I could tell you why I'm the right person to build this. But the people who worked alongside me can tell you better.
“Brian is my first boss and has been an invaluable mentor as I've entered the corporate world.”
— An engineer I mentored through his first corporate role
“He didn't just manage outcomes. He identified potential, nurtured it, and consistently challenged me to think bigger, act sharper, and grow beyond my perceived limits.”
— A data and reporting leader on my team
“Brian is a true leader who inspires rather than manages, leading his people with vision and purpose. He genuinely cares about those he leads, not only in their work but also in their well-being.”
— Our Agile coach
“Brian acted as a constant shield for our team, absorbing management escalations and corporate pressure so we could focus entirely on innovation.”
— A senior engineer I recruited to the team
Their full recommendations are on my LinkedIn profile.
Why it's free for early-career professionals
Career Vault is free for early-career professionals. That's a founding decision, not a promotion.
When I was building my career, nobody handed me a framework. Nobody told me what a career was supposed to look like six waves in, or how to think about the difference between building toward something and drifting through promotions. I don't want anyone else to wait that long.
The math works: early-career users who build the habit of documenting their career become mid-career professionals who see the value and subscribe. They fund the next cohort. The product sustains itself by investing in the people who need it first.
Where this is going
I've been on the other side of career services. I once paid $800 for a resume revamp and brand assessment. I walked away halfway through because it was clearly going to be a waste of time. There were no refunds. That's not a knock on the profession. There are career coaches and resume strategists who genuinely transform careers. The problem is access. Those services cost hundreds or thousands of dollars, and they're largely available only to executives with disposable income.
Career Vault exists to change that equation. Structure your career data once, verify it once, and build a foundation that makes every career service more effective, more affordable, and more accessible. My goal isn't to automate what great career professionals do. It's to make sure you have options, and that the tools to manage your career aren't reserved for the people who need them least.
What I won't compromise on
These are personal commitments. If I break any of them, the product is no longer mine.
Your data stays yours.
No recruiter access to your vault. No selling career data to third parties. No monetizing your professional history without your explicit consent. I spent a decade as the person federal regulators trusted with data on millions of consumers. I'm not going to build a product that treats your career data with less respect.
Truth before generation.
AI in Career Vault will never fabricate, inflate, or embellish your career record. If you didn't do it, the system won't say you did. I held an enterprise AI model to under 0.5% material hallucination by grounding every output on verified source data. I apply the same standard here.
No dark patterns.
Free means free. No trial clocks. No feature walls that make the free version useless.
Governance by design, not afterthought.
Every system I built in my career had controls designed in from day one. Career Vault follows the same principle.
I will kill features that don't meet the standard.
I've done it before. It cost me political capital and timelines. I'll do it again. If a feature can't be trusted, it doesn't ship.
Career Vault is early. The product is evolving. Some ideas I'm excited about today won't survive contact with real users. But the architecture won't change: structured data first, verified by the person who lived it, owned by the person who created it. Everything else gets built on top of a foundation I trust.
I'm sharing this before the product is finished because I believe the people who will use it deserve to know who built it and why. I'd rather earn your trust now than ask for it later.
If you want to know more about me, my LinkedIn is here. If you want to reach me, here.
Career Vault is small right now. But it's honest. And in my experience, honest scales.
