BACK TO ALL NEWS & INSIGHTSEvan Robinson’s career has been shaped by a constant pursuit of speed, efficiency, and elegant systems. From his early days engineering scalable platforms at Yahoo to his tenure at Shipwire, where he helped pioneer e-commerce fulfillment technology, Evan has always been drawn to the intersection of digital precision and physical complexity.
As Jitsu’s Chief Technology Officer, he’s channeling that experience into the next evolution of last-mile delivery. Under his leadership, Jitsu’s technology is built not just to keep up with demand, but to anticipate it—leveraging predictive analytics, intelligent automation, and flexible architecture to make reliability a competitive advantage.
Read our full conversation with Evan below.
What originally drew you to logistics, and what challenges kept you hooked?
Okay, story time. I didn’t get into logistics until about ten years into my career, and if you’d told me back then that shipping would become the most rewarding part of my professional life, I’d have looked at you funny. But that’s how it played out.
The story starts in 2006. I was at Yahoo working on Yahoo Autos—a dream gig for a gearhead like me. Yahoo was still the site on the Internet back then, and before the era of “cloud everything,” scaling a web app meant buying and tuning physical servers. Developers had to understand the whole stack—front-end, back-end, ops, databases—because you couldn’t just throw more hardware at a problem without a pretty good reason. You learned to make things go faster, and you learned it fast.
Yahoo, however, was starting to slow down organizationally and technologically. They were shifting from being a technology pioneer to a media company, and that pace didn’t suit me. I don’t like slow—on the road or in my code.
Around that time, my brother-in-law, Damon Schechter, was experimenting with a startup idea in e-commerce fulfillment. He’d founded something called Shipwire—basically “Hotwire for shipping” if you remember that travel site. He built version 1.0 himself while lining up the warehouse partnerships, handling sales, and writing the app. He was getting traction but needed someone to take the technology reins. He saw how restless I was at Yahoo and made me an offer: “Come on board, but listen, l never want to have to worry about code again.” We shook on it. That handshake changed everything.
Shipwire was where I fell in love with logistics. Three things stood out immediately:
First, the challenge of solving problems at the intersection of the digital and the physical. You can test software, debug it, and know when it’s working. But the human factor—warehouse workers, carriers, and customers—throws beautiful chaos into the mix.
Second, mastering the art of doing more with less. Logistics is a high-volume, low-margin game. It rewards efficiency and optimization—the same things I loved about tuning cars and code.
And third, building flexibly but neatly. Growth is great, customer-focused is great, but there’s a fine line between being customer-driven and building a spaghetti system. There’s an art to saying yes without creating a mess.
Shipwire grew steadily. We learned to make fulfillment technology that didn’t just move boxes but understood client needs in real time. Seven years later, we sold to Ingram Micro. It’s still going strong today (now under CEVA), and I’m proud that a few Shipwire alumni—folks like Chantra Park and Sampson Wu—are now here with me at Jitsu.
After Shipwire, I joined Zesty.ai, a team of brilliant ML and computer vision engineers modeling wildfire risk for the insurance industry. I learned a lot, but I also learned what I wasn’t. Their hardest challenges were deep learning problems, and they were very well-staffed to tackle those; my stock in trade had always been systems and scaling, and we’d gotten those things to a good place during my tenure there. So when a recruiter called about an up-and-coming delivery start-up called Jitsu, I was intrigued.
At first, I hesitated. I’d just spent a decade in logistics and wanted to keep things fresh. But when I saw the demo, it felt like the early days at Shipwire—this vast, greenfield landscape of problems waiting to be solved. Except this time, the challenges weren’t inside the warehouse—they were on roads and doorsteps. How do you orchestrate a delivery experience so seamless that a brand new driver can perform like a seasoned pro? How do you make routing, pricing, and sortation smarter every single day?
From an optimization standpoint, these were endlessly tall trees to climb. And from a human standpoint, there were puzzles abound in customer empathy and design. Best of all, the whole system felt alive—constantly moving, generating data, logging outcomes you could learn from and refine.
So, what started as a love of making cars and code go faster turned into a love of helping cars and couriers deliver faster (and more reliably). And I guess trucks too.
Which technologies do you believe will most transform last-mile operations, and how is Jitsu positioning itself to leverage them?
Logistics sounds simple: get something from point A to point B. But the real craft is in the last 1–2%—that final stretch of reliability. Everyone can hit 96–98% on-time delivery. Getting to 99%+ takes relentless, almost obsessive exception management.
That’s where predictive analytics comes in. Traditional logistics planning is static—plan your volume, schedule your fleet, cross your fingers. But real life doesn’t stick to plan. Trucks arrive late. Drivers call out sick. Clients drop surprise promotions. Predictive systems help us anticipate these issues and adapt before they happen.
Early on, we did what a lot of startups do—we threw talented people at problems. When a delivery got messy, a human fixed it. That worked when we were small and scrappy, but heroism doesn’t scale. More recently, we’ve started asking smarter questions: Was that extra volume really unpredictable? Could we have seen that pattern coming? Now, we’re feeding those learnings back into our planning models.
The other frontier I’m excited about is agentic tech—AI systems that can act on your behalf in complex, changing environments. We serve four groups: clients, their recipients, drivers, and our internal teams. When “stuff happens”—a delivery reroute, a missing access code, even a software bug—those four groups all need help. Historically, that meant tickets and chat queues. Now, smart agents can triage known issues, guide users through them, and free humans for the real curveballs.
But agents aren’t a silver bullet. If someone “has to talk to the manager,” even if that manager is a fast, helpful AI, it means we didn’t design some workflow well enough. We see agents as companions to good design, not a replacement for it. The goal isn’t to eliminate human involvement, it’s to make every human moment count.
What core principles guide Jitsu’s tech architecture?
Three big ones.
First, as I mentioned before, build flexibly but neatly. Innovation and technical debt go hand in hand, but there’s a difference between strategic shortcuts and structural rot. If you cut a corner, do it knowingly, and make sure you come back to fix it. Otherwise, you’re laying landmines for future teams. I’ve worked on those “Winchester Mystery House” systems before—rooms and hallways built on top of each other with no plan. It’s no fun.
Second, keep it simple until it can’t be. Delivery is deceptively complex, with a small vocabulary—clients, shipments, routes, drivers, warehouses—but endless variations. You can’t change the domain but you can make it more pleasant for a new engineer to work in. When I joined Jitsu, I was impressed because the system was elaborate but remarkably consistent. As we scaled from dozens to hundreds of microservices, we kept that regularity. Every service feels familiar. Like walking into a Courtyard by Marriott—even if you haven’t visited a particular one, you kind of know what to expect and where everything is.
Third, plan for failure and recovery. Every year, we see a few major cloud or carrier outages. Reliability doesn’t mean nothing breaks, it means you’re ready when it does. We build fallback systems, run tabletop and live drills, and make sure even the low-tech contingencies work. I’ve literally seen days where printing routes on paper was the difference between chaos and continuity. Reliability sometimes comes down to toner.
And finally, combine problems with passions. People do their best work when what they love intersects with what the company and your customers need. The most exciting innovations rarely start as roadmap projects—they start as experiments, side tinkering, or someone scratching an itch. You have to leave space for that.
What recent tech project at Jitsu has delivered the most impact for clients or internal teams?
Two come to mind: our Client Control Tower Dashboard and a new service offering.
The Control Tower is all about visibility. Before you can move the needle, you have to see it. We’ve always made it easy for clients to export performance data, but that wasn’t enough. The dashboard now gives a holistic network view—on-time rates, first-mile accuracy, SMS effectiveness, geofence precision—all surfaced in one clean, intuitive interface. It’s not just data for data’s sake; it’s insight they can act on.
Our new service offering was a fun systems and operational challenge. We wanted to offer a different delivery option with the same reliability and transparency Jitsu is known for. To make that work, we rebuilt our inventory system, redesigned our sortation processes, and parallelized our facility operations to handle multiple cycles simultaneously. It was a major effort, but it unlocked a whole new segment of clients and made our internal operations far more scalable.
As last-mile networks become increasingly data-driven, how do you think about using data not just for visibility, but for predictive decision-making and optimization?
We think prediction and proaction are the next great frontier. Visibility tells you what happened. Prediction helps you prevent it from happening again—or better yet, use it to your advantage.
At Jitsu, every data point—shipment volume, route performance, driver reliability, delivery issues—feeds into models that help us adapt before things go wrong. The real shift, though, is cultural. You have to train teams to think probabilistically, not deterministically. Instead of “what does our forecast say we should expect,” we ask “what range of outcomes could we see, and how do we prepare for each?” Once you do that, everything from staffing to routing becomes smarter and less stressful.
Looking ahead, what’s the next big turning point in last-mile tech?
At the risk of repeating myself, it’s adaptive orchestration—systems that anticipate, decide, and then react as needed in real time across the entire network, with human intervention only for outlying anomalies. Companies that can more accurately forecast inbound volume, delivery risk, and labor supply simultaneously will outpace everyone else.
The winners will automate what’s predictable and reserve the need for human judgment for the truly exceptional situations.
On a personal note, how do you recharge or stay creative outside of work?
When you think about the same problems every day, tunnel vision creeps in. My reset button is changing the scenery—a new coffee place, a park I don’t visit too often, a bench where people pass by. The goal is to surround myself with things I don’t see every day. New spaces bring new ideas.
And my best mornings are the ones I don’t start in reactive mode. A couple of times a week, I’ll skip Slack and email, make coffee, and just play—reading, sketching ideas, or coding a prototype for something half-baked in my head. Those mornings always turn into the most productive days.
In the end, Jitsu’s mission—and mine—isn’t just to move packages faster. It’s to keep evolving how the whole system moves: faster, smarter, and, if we do it right, a little more beautifully every week.