Enova International

Enova is hiring for a Principal Software Engineer - Platform (hybrid) in Chicago! Please reach out if you are interested or apply here! https://lnkd.in/guzdsA-v

Do you want to work someplace awesome? Enova is hiring for a Collections Representative (Remote). Apply here https://lnkd.in/gWC9TcnG .

I looked back at my Q1 meetings last week. I spent 80-90% of my day in meetings. Pretended I was listening. What I was actually doing: queuing up my next sentence while the other person was still talking. Waiting for the gap so I could land my point. Calling that "engaged." A class at Kellogg made me try one small thing. Wait five seconds after someone finishes before you respond. That's it. The first week was awkward. Five seconds in a meeting feels like a long time. I felt the urge to fill it every single time. Two things changed. I stopped talking over people, which I hadn't realized I was doing as much as I was. And the pause forced a thought instead of a reaction. The response I gave after five seconds was almost always better than the one I would've fired off. The mentors I respected most all did this. I never clocked it because they made it look like composure. They were just pausing. Five seconds. Still the hardest part of my day. #Leadership #ActiveListening #KelloggMBA

About a year ago I took on a new challenge at work, and for a while it pretty much ate me alive. Not in the dramatic way. In the quieter way where you notice you haven't picked up a book in two months and your evenings have become a blur of doom scrolling because that's all the bandwidth you have left. Part of what made it hard was inheriting a project that had been partially overlooked for a while. By the time I got there it was already a sore spot — for stakeholders, for the team, for people who'd been watching it drift. Every piece of criticism that came in, I took personally. Which is objectively ridiculous, because I wasn't even around when most of it was set in motion. Didn't matter. My brain had decided this was mine now, and that meant every bit of frustration aimed at the project was aimed at me. I'd been carrying this around for months when I got on a 60-minute call with a career coach I've worked with on and off for about a decade. I was venting. He let me. Then he said one sentence that honestly rewired how I was looking at the whole thing. "You're just here to help. You didn't cause the problem but you're going to solve it." That was it. That was the whole thing. Did I personally solve it? No. But I think I helped pave the way for the team to solve it — and we've just closed out two strong quarters back to back. I'm not going to pretend that's all because of one sentence from a coach on a Tuesday afternoon. But something shifted when I stopped carrying it the old way, and the team felt it too. Turns out that's actually the job. Not to absorb the weight of every past decision personally, but to clear the path so the people around you can do the work. If you're in the middle of something heavy right now that predates you — that one's for you too.

I spent the last month building an agentic team to ship a real project — more to come on that in the future — mostly as a way to actually learn AI instead of reading about it. The thing that surprised me: almost none of the hard work was about the agents themselves. I assumed the challenge would be prompting, designing the right specialists, getting the model to write good code. It wasn't. The hard part was the scaffolding around them. Permission checks before delegation (an agent without write access will still cheerfully attempt to write files and fail). Fallback paths when a specialist stalls after two attempts. An attribution log so when something broke, I could actually tell which agent did it and why. A shared memory directory so each new session didn't start from scratch re-reading the codebase. None of this is the stuff that shows up in demos. It looks a lot more like the plumbing you build around a flaky distributed system. What I took from it: agentic systems fail in the same places regular engineering systems fail. - Coordination, - Observability, - Graceful degradation. The models are already good enough for real work. The engineering now is making them reliable enough to trust with something that matters. If you're thinking about building with agents, I'd spend less energy picking the fanciest model or architecture and more time thinking about how you'll know when one of them quietly does the wrong thing. #EngineeringLeadership #AgenticAI #AI #SoftwareEngineering

A CEO just bought $2M of their own company's stock at a 52-week low. No press release. No analyst upgrade. Just a Form 4 filed with the SEC. This happens more than you'd think. And it's one of the most underused signals in public markets. Here's why: insider buying is asymmetric information made legal. These people see the quarterly numbers, the pipeline, the board discussions. When they spend their own money — not options, not RSUs, actual cash — it means something. The academic research backs this up. Studies going back decades show insider purchases outperform the market on average. The catch? There are hundreds of these filings every day, most of them noise (option exercises, gifts, automatic plans). Finding the signal in the noise is the whole game. That's why I built Buyside Brief. It scores every filing on 12 signals and sends the best ones to your inbox before the market opens. Free, every weekday.

For a while my AI workflow for trading looked embarrassingly manual. Export positions from tastytrade. Paste into Claude. Ask a question. Realize I needed chain data too. Go back. Pull the chain. Paste that. Repeat for IBKR. By the time I had everything in context I'd forgotten what I was trying to figure out. The analysis was good. The friction was killing it. So I spent a weekend fixing the pipe instead of the prompt. For tastytrade I stood up an MCP server so Claude can pull positions and option chains on demand — ask about a ticker, get the live chain, no copy-paste. IBKR was less cooperative, so I wrapped some Python around their gateway and dumped the data to JSON that feeds into Claude directly. Ugly but it works. Then I built a small Vite/React app I've been calling Alpha Lab to hold the rules and the paper-trading challenges — versioned guidelines, position caps, stop-out triggers, a closed-trades ledger. The rules live in the app. The data comes from the brokerages. Claude sits in the middle and does the analysis against the rules. What changed wasn't the quality of any single answer. It was the frequency. Audits went from "once a quarter when I felt guilty" to "whenever a question occurred to me." First thing that surfaced: a covered call position that had quietly stopped working. Stock had dropped well below my short strike months earlier, the call premium was pennies, and the position was dead money pretending to be an income trade. I knew. I just hadn't wanted to do the work to confirm it. Took Claude about thirty seconds once the data was wired in. The thing I keep telling engineers on my team and haven't quite internalized for myself: the leverage in AI workflows isn't in better prompts. It's in removing the steps between question and answer. A mediocre model with zero friction beats a great model with five manual handoffs, because you'll actually use the mediocre one. Most people's AI workflows die in the copy-paste. What's a workflow you've been prompting your way through that should probably be wired up instead?

Career conversations don't have to be rough. But the honest ones usually are. An engineer on my team once asked me why they hadn't been promoted. They were hitting every expectation of their current role. Shipping on time. No issues. Reliable. I didn't have a great answer. Not because they were wrong — they were doing their job well. But doing your current job well is the baseline. It's not the case for the next one. That conversation taught me something about what a manager actually owes you — and what they don't. Your manager can't build your career for you. They can't want your promotion more than you do. They can't chase down opportunities and hand them to you and then make sure you follow through. That's your job. But here's what a good manager does: They sponsor you in rooms you're not in. When your name comes up in a calibration or a staffing decision, they make sure people know what you're capable of — not just what you've been assigned. They create opportunities but don't chase them down for you. They'll put you in front of the work that could stretch you. Whether you grab it is on you. They know when to push you and when to back off. Some people need a nudge. Some need space. A good manager reads that and adjusts. If you're waiting for your manager to build your career, you'll be waiting a long time. But if you're driving and your manager isn't doing those three things — that's worth a conversation too. It goes both ways. What's the best thing a manager ever did for your career? #engineeringleadership #management #careergrowth #softwaredevelopment #leadership

Time flies when you're having fun. Incredibly proud of what we've accomplished with the NetCredit DM program over these last 2 years. Huge thanks to Brad Nelsen for the trust and the constant push to keep leveling up. Looking forward to continuing the momentum in the years ahead! #InsideEnova