Founder Notes
From Google ML to Hemp Delivery: Why I Left Google to Build Compliant Hemp Delivery in Texas
From Google ML to Hemp Delivery: Why I Left Big Tech for a Regulated Industry
The Decision That Confused Everyone
In February 2024, I walked away from a career trajectory that most engineers would envy. I'd spent years at Google building machine learning systems—the kind of work that appears in research papers and powers products used by billions. My LinkedIn was pristine: Google, AI/ML, five Udacity certifications including Deep Learning and Reinforcement Learning nanodegrees, leadership roles. I was on the path.
Then I quit to build hemp delivery software in Texas.
Google ML, voice assistants, and AI infrastructure before pivoting to regulated industries
The reactions were predictable. "Hemp? Like... marijuana?" No, hemp—the legal variant with ≤0.3% THC. "In Texas? Isn't that a legal gray area?" It was, which is exactly why the opportunity existed. "Why would you leave Google for that?" Because the impact-per-line-of-code ratio was dramatically higher.
At Google, I was one of thousands of talented engineers working on problems that millions of other talented engineers could also solve. At HempDash, I'm applying AI and software engineering skills to an industry that desperately needs them but that most tech talent refuses to touch. The competitive landscape is different when your competitors are Excel spreadsheets and handshake deals.
The Career Path That Led Here
My path to HempDash wasn't random. It was the accumulation of skills and experiences that turned out to be perfect preparation for building a regulated delivery marketplace—I just didn't know it at the time.
The Google Years: Scale and Systems Thinking
At Google, I learned what "scale" actually means. Not "a lot of users" scale, but "one bug can cost millions of dollars" scale. I built ML systems that had to be right, every time, at massive throughput. I learned to think in systems—how components interact, where failures cascade, what observability you need to sleep at night.
These skills transferred directly to HempDash. A marketplace isn't that different from an ML pipeline: inputs (orders), transformations (routing, fulfillment, verification), outputs (deliveries), and a whole lot of edge cases that can blow up if you're not careful.
Voice Assistants and Real-Time Systems
Before HempDash, I co-founded Lailia AI, building voice assistant technology. That experience taught me about real-time systems, user experience design for non-technical users, and the brutal reality of startup fundraising. Lailia didn't become a unicorn, but it gave me the founder immune system I'd need for HempDash.
The Pattern: Applying AI to Underserved Industries
Looking back, my career has been about taking AI/ML capabilities and applying them to domains that haven't been touched by Silicon Valley. Voice assistants for everyday users. Delivery optimization for a regulated industry. The common thread is finding places where technical sophistication creates disproportionate value because the baseline is so low.
Why Regulated Industries Need Tech Founders
Here's the uncomfortable truth about the cannabis/hemp industry: it's running on 1990s technology. I've talked to dispensary owners who track inventory in Excel. Delivery services that use WhatsApp for driver coordination. Compliance documentation stored in filing cabinets.
Growing industry with limited technology competition—most tech talent avoids regulated spaces
This isn't because industry participants are unsophisticated. It's because the tech talent that could build better solutions is busy chasing adtech, SaaS, and crypto. Regulated industries are seen as "too complicated" or "too risky" by most software engineers. That perception gap is the opportunity.
The talent arbitrage is real. A mediocre engineer at a FAANG company is surrounded by exceptional engineers doing similar work. A competent engineer in cannabis tech is often the most sophisticated technical person in the room. The same skills that make you "good" at Google make you "exceptional" in underserved industries.
Regulatory complexity creates moats. At Google, any competitive advantage I built could be replicated by competitors with similar resources. In hemp delivery, the 5-7 months I spent building compliance infrastructure (age verification, COA tracking, audit trails) is a moat that new entrants have to match before they can compete. Complexity that scares away competitors is your friend.
Impact per line of code is higher. At scale, a Google engineer might improve a metric by 0.1% and that's considered a success because 0.1% of billions is still a lot. At HempDash, I shipped features that took vendors from "can't legally operate" to "fully compliant" overnight. The before/after is dramatic and visible.
The Skills That Transferred (And The Ones I Had to Learn)
Not everything from big tech translated to startup life. Here's an honest breakdown:
What Transferred Directly
System design and architecture. Thinking about how components interact, where to put boundaries, how to handle failures—this transfers completely. HempDash's backend architecture looks like something I would have designed at Google, just smaller.
ML/AI for logistics. Demand prediction, route optimization, recommendation engines—these are core AI problems I'd worked on before. The domain is different (hemp products instead of search results), but the underlying techniques are identical.
Data pipeline thinking. Compliance requires moving data from point A to point B reliably: orders, verifications, audit events. This is just ETL with legal consequences. My experience building data pipelines at scale was directly applicable.
Observability and debugging. The instinct to instrument everything, log comprehensively, and build dashboards was invaluable. When you're a solo founder debugging a production issue at 2 AM, good observability is the difference between a 30-minute fix and a 4-day debugging nightmare.
What I Had to Learn
Regulatory thinking. At Google, "move fast and break things" was aspirational. In regulated industries, "move carefully and document everything" is survival. I had to develop a new instinct: before building any feature, ask "what are the compliance implications?"
1.3% of age verification checks failed due to customer actually being under 21
Stakeholder management complexity. At Google, my stakeholders were other engineers and product managers. At HempDash, I work with vendors (small business owners with varying technical sophistication), couriers (gig workers with different expectations), customers (consumers who just want their products), and regulators (government officials with enforcement authority). Each group needs different communication.
Patience with slow-moving industries. Tech moves fast. Regulated industries move slow. Vendor onboarding that takes 2 weeks in a SaaS business takes 2 months when you need business licenses, insurance verification, and compliance documentation. I had to recalibrate my expectations.
Comfort with ambiguity. Startup founders always talk about ambiguity, but regulated industry ambiguity is different. It's not "we don't know if users will like this feature" ambiguity—it's "we don't know if this feature is legal in 6 months" ambiguity. You have to build systems that can adapt to regulatory changes you can't predict.
The Case for Leaving Big Tech
I'm not here to tell everyone to quit their FAANG job. Big tech is a great place to build skills, earn good money, and work on interesting problems. But if you're a senior engineer wondering "what's next?", here's the case for regulated industries:
1. The skills premium is higher.
The same abilities that make you a "senior engineer" at Google make you a "technical co-founder" in underserved industries. Your skills are rarer and therefore more valuable. You can build an entire company instead of building 0.1% of a product.
2. Ownership is real.
At Google, I "owned" projects that the company could kill at any time. At HempDash, I own equity in something I'm building. The upside is uncapped, but more importantly, the connection between effort and outcome is direct and visible.
3. Impact is tangible.
I've met vendors who tell me HempDash let them quit their day jobs and build real businesses. I've talked to customers in rural areas who finally have access to products that help them. The impact is concrete in a way that "we improved click-through rate by 2%" never was.
4. The market timing is favorable.
Cannabis/hemp is a $30B+ industry growing 15%+ annually, with limited technology competition. The regulatory environment is complex but stabilizing. The window for building foundational infrastructure is now—in 10 years, the market leaders will already be entrenched.
5. Purpose matters.
This is subjective, but for me: I care more about building compliant delivery infrastructure that helps small businesses than about optimizing ad targeting. Your values may differ, but if "what you're building" matters to you, regulated industries often offer more meaningful work than consumer tech.
What I'd Tell My Past Self
If I could talk to the version of me who was considering this leap, here's what I'd say:
The transition is harder than you think, but also easier than you fear. The learning curve for regulated industries is real—compliance thinking doesn't come naturally to tech-trained brains. But your technical skills transfer more directly than you expect. You're not starting from zero; you're starting from a strong foundation and adding new layers.
Your network won't understand, and that's okay. Tech friends will think you're crazy. Finance friends will ask if hemp is legal. Family will be confused. You'll have to explain your company 50 times before your pitch is crisp. This is normal.
The loneliness is real but manageable. Going from a team of 50 engineers to solo founder is an emotional shift. Find founder communities (Y Combinator's alumni network, local startup groups, industry associations). The technical problems are solvable; the isolation is what breaks people.
Move faster on the basics, slower on the novel. I wasted time building custom solutions for solved problems (should have just used Auth0 from day one). I also moved too fast on compliance infrastructure and had to rebuild parts. The rule: use off-the-shelf for commodity features, invest time in what differentiates you.
Trust the process. There will be moments—many moments—when you wonder if you made a mistake. When a key vendor churns, when a regulatory change threatens your business, when you're debugging at 2 AM and missing your old life. These moments pass. The process works if you stick with it.
Where This Story Goes Next
Eighteen months after leaving Google, HempDash is live in Texas with 50+ vendor partners, zero compliance violations, and a technical infrastructure that took 10 months to build solo. We're preparing for expansion to Oklahoma and Louisiana. We're hiring our first engineers.
The decision to leave big tech for hemp delivery was unconventional. It confused my former colleagues, worried my family, and required betting my career on an industry most tech talent avoids. But it also let me build something from scratch, apply my skills where they're rare and valuable, and have tangible impact on real businesses and customers.
If you're a senior engineer at a big tech company, wondering if there's something more—there is. It might not be hemp delivery. But somewhere, there's an industry running on Excel spreadsheets, waiting for someone with your skills to build something better.
The question is whether you're willing to leave the comfortable path to find it.
Frequently Asked Questions
Is hemp delivery actually legal in Texas?
Yes. HB 1325 (2019) legalized hemp and hemp-derived products in Texas, and SB 3 (2023) established the regulatory framework for sales. Hemp is defined as cannabis with ≤0.3% Delta-9 THC by dry weight. We operate fully within this legal framework with age verification, COA tracking, and compliance with Texas DSHS requirements.
How do AI/ML skills from big tech apply to delivery logistics?
Directly. Demand prediction (when will orders come in?), route optimization (how should drivers navigate?), recommendation engines (what products should we suggest?), and fraud detection (which orders are suspicious?)—these are all ML problems I worked on at Google, applied to a different domain. The techniques transfer; only the data and business context change.
What's the biggest mistake tech founders make when entering regulated industries?
Building first, researching regulations later. In consumer tech, you can ship fast and fix compliance issues as they arise. In regulated industries, a compliance mistake can mean business license revocation or legal liability. The correct order is: understand regulations → design compliant architecture → build features within that architecture.
How do you evaluate if a regulated industry opportunity is worth pursuing?
Look for three things: (1) Is the industry large enough to support a real business? (2) Is current technology usage significantly behind what's possible? (3) Is the regulatory environment stable enough to build on, or still in flux? Hemp/cannabis checks all three: $30B+ market, running on Excel, and regulations are complex but stabilizing.
Do you regret leaving Google?
No. I miss certain things—the resources, the colleagues, the intellectual environment. But I don't miss the limited ownership, the disconnection from impact, or the sense that I was one of thousands of interchangeable engineers. HempDash is harder, scarier, and more rewarding. I'd make the same choice again.
About Jonathan Sullivan
Jonathan is the founder and CEO of HempDash, a compliant hemp delivery marketplace operating in Texas. Prior to HempDash, he spent 12+ years in tech including roles at Google building machine learning infrastructure and co-founding Lailia AI. He studied Mathematics and Computer Engineering at the University of North Texas and holds five Udacity certifications in AI/ML. Connect with him on LinkedIn or Twitter/X.
_If you're considering a similar career transition—from big tech to regulated industries—I'd love to hear your story. The path is unconventional but the opportunity is real. Reach out on LinkedIn or Twitter._