I Resold Model Trains and Electronics to Pay My Server Bills. Here's What Changed Everything.

The Infrastructure Problem: $500–800/Month Just to Stay Alive

Building Botifex involved three expensive things. Combined, that was $500–800/month just to keep the lights on:

• Servers and Hosting — Botifex needs to monitor six marketplaces 24/7. That's not a shared hosting situation. That's dedicated infrastructure. Cloud services. Databases that scale. Bill: ~$250–300/month.
• Databases — Monitoring Craigslist, eBay, Facebook Marketplace, OfferUp, Poshmark, and Mercari simultaneously means storing millions of listings. Real-time data. Historical data for price tracking. Bill: ~$150–200/month.
• AI Model Training — The Deal Score feature—the core value of Botifex—runs on machine learning models trained to recognize deals. Training, hosting, and running inference on those models isn't free. Bill: ~$100–150/month.

The Reselling Side Hustle: $2,000–5,000/Month to Fund Infrastructure

At 16, even with part-time job prospects, $500–800/month isn't realistic. I needed that much in revenue just to cover infrastructure. Any additional revenue would go to continued development, testing, or reinvestment. So I started flipping.

I had a genuine interest in model trains—collecting, restoring, reselling them. That interest became a starting point.

But I quickly realized: I couldn't scale on model trains alone. The market was too niche. Instead, I flipped what the market offered: electronics, furniture, vintage collectibles, anything with margin.

I set a goal: $2,000–5,000/month. That would cover all infrastructure costs, reinvestment, and give me bandwidth to keep building.

Here's the crucial part: I built this entire operation using Botifex.

I wasn't building Botifex in a vacuum and then using it casually. I was using it obsessively—every single day—to source deals. Every problem I hit as a reseller became a feature request to myself as a developer.

Feature #1: Deal Score (Born From a $90 Mistake)

Three weeks into reselling, I bought a MacBook Pro.

Listing price: $400. I'd done a quick eBay search and saw prices around $650–750, so I bought.

Got home. Listed it. First serious interest: $420 offer.

I almost took it. Then I looked deeper into the sold listings. The model was older than I'd realized. Actual market value? Closer to $500–550. I'd be taking a $90 loss on this deal.

At $500–800/month in infrastructure costs, that $90 mistake was 15% of my monthly budget.

Never again.

That night, I built Deal Score into Botifex.

The feature pulls real eBay sold listings for the exact item (not asking prices, not current listings—actual sold prices), compares your buy price to the market average, and gives you a 0–100 score.

• 0–30: Don't buy.
• 30–60: Okay deal.
• 60–80: Good deal.
• 80–100: Excellent deal.

I immediately tested it. Found a Sony camera listed at $600. Deal Score: 82. Average market price: $950–1,050. Bought it. Listed it. Sold for $1,000 five days later.

That $400 profit covered my database costs for the month.

Feature #2: Real-Time Multi-Marketplace Alerts (Born From Missing Deals)

By the end of week two, I realized something painful: good deals appeared everywhere, not just Craigslist.

Facebook Marketplace had solid local sourcing. eBay had auctions closing with soft competition. OfferUp was picking up volume. Poshmark for clothing. Mercari for collectibles.

But I could only monitor Craigslist with my scraper. Everything else meant manual refreshing.

If I spent 30 minutes refreshing eBay, I'd miss the Facebook deal. If I checked OfferUp, I'd lose the Craigslist window. Other resellers were posting about deals they'd found while I was busy monitoring a different platform.

This wasn't just inefficient. It was costing money. Deals I could have won were going to other resellers who were monitoring more platforms.

Solution: Expand Botifex to scrape all six marketplaces simultaneously. Not one marketplace with occasional manual checking on others. Real alerts across all six the moment a deal matched my criteria.

But here's the problem: adding more marketplace scraping to the Raspberry Pi accelerated its death. Five scrapers running meant constant crashes. Six scrapers meant the Pi couldn't keep up.

That's partly what forced the infrastructure upgrade. By the time I moved to real servers, I built out all six marketplace scrapers at once. Suddenly, deal velocity doubled. Instead of 10–15 potential sourcing opportunities per week (most from Craigslist, some manual checking elsewhere), I was seeing 20–30 from all six platforms at once.

Same conversion rate (about 30–40%), but higher volume meant more inventory cycling. More inventory cycling meant more profit velocity.

The math: 15 deals per week times 30 percent conversion times $200 average profit equals about $900/week. Double the deal flow: $1,800/week. That's $3,600/month versus $7,200/month revenue.

I was now clearly above my infrastructure costs. I could keep building.

Feature #3: SMS and Push Alerts (Born From Sleeping Through $300 Profit)

This one still stings.

Around month three, I set an alert: MacBook Air M1, max price $500. A deal hit at 2:47 AM.

I was asleep.

At 7 AM, I checked Botifex casually (before school). The listing was four hours old and already marked sold.

I'd missed a $300+ profit because the deal moved while I was unconscious.

That day, I built SMS alerts into Botifex. Real text messages. Not "check your dashboard later"—actual notifications that wake you up.

The next deal came in at 1:15 AM on a Wednesday. SMS hit my phone immediately. I woke up, messaged the seller within 90 seconds, bought it, resold it three days later for $310 profit.

That 90-second response window was the difference between winning and losing.

Feature #4: Deal Price Analytics (Born From Listing Confusion)

By month four, I had inventory in various stages: some photographed, some listed, some sitting in a box waiting for photos.

When I listed something, I'd waste time trying to price it optimally. Check eBay sold listings, check Poshmark, check Facebook—all for one item. Thirty minutes per listing.

Meanwhile, inventory sits unsold. Unsold inventory means tied-up capital. Capital locked in one item means I can't source the next deal.

Solution: Price analytics integrated into Botifex.

When you source a deal, you see immediately:

• eBay sold listings for the exact model/condition
• Regional price variations
• Average time to sell
• Current trend (prices rising or falling)

Now I list in 10 minutes instead of 30. Faster listing means faster sales. Faster sales mean capital cycling quicker, which means I can source more volume.

One camera that sat for a week (while I was guessing at price) now sells in 48 hours (with data-backed pricing). That's the difference between $3,600/month and $5,000/month revenue because capital cycles faster.

Feature #5: Inventory Management (Born From Complete Chaos)

Month four: I had eight items in various stages of the resale pipeline.

One was photographed and listed. One was photographed but not yet listed. One was sitting unphotographed. One had sold but wasn't shipped yet. One was shipped and waiting for delivery confirmation.

I had a Google Doc tracking all this.

One day, I almost re-listed an item that had already sold because my Google Doc was out of sync with reality.

That's when I realized: I need an inventory system inside Botifex itself.

Built a simple tracker: upload photos, note condition and price, mark status (sourced, photographed, listed, sold, shipped). One source of truth.

Sounds basic. But the difference between chaotic reselling and scalable reselling is systems. Once I had a clean inventory system, I could run 20–30 simultaneous items instead of 5–8. The operational ceiling went up.

The Inflection Point: Month Six

Around month six, something shifted.

The reselling operation was stable at $4,000–5,000/month. My infrastructure costs were easily covered. I had credibility—I could point to early users and say, "I use this tool every day to resell. Here's exactly what I found."

More importantly: the product was generating value that clearly exceeded my own use case.

Users were signing up who had nothing to do with model trains or my personal sourcing strategy. They had their own categories, their own workflows. But Botifex solved their problems because every feature was rooted in real reseller pain.

At that point, I could have stopped reselling. The infrastructure was self-funding.

But I didn't.

I scaled back to 5–10 high-margin flips per month (mostly just testing new feature ideas, or sourcing projects that validated whether a new tool would actually help resellers). Not for revenue. For empathy.

Because staying close to actual reseller problems meant I'd never build features based on theory. Every feature request would go through the filter: "Would I use this if I were reselling?"

That filter is why Botifex doesn't have unnecessary complexity. Every feature exists because it solves a real problem that real resellers face.

What I Learned From Funding Infrastructure Through Reselling

Five lessons that stuck:

• Use your own product obsessively. I wasn't building Botifex in isolation and occasionally testing it. I was using it every day to source deals, price inventory, and manage sales. That daily use meant I found problems quickly.
• Your infrastructure costs are real, and they matter. $500–800/month doesn't sound like much until it's your problem. At that price point, you need recurring revenue. You need a real business model, not a side project. That constraint forces you to build something people actually want.
• Stay close to customers even after you don't need to. The second I could stop reselling, I could have gone full-time product. But staying in the market—doing a few flips every month—means I catch problems that users mention casually but haven't formalized into feature requests. That's where real insights come from.
• Features born from personal pain are features that ship. I didn't theorize about features. I lived through problems, built fixes, and shipped them immediately. That velocity meant Botifex improved faster than if I'd been guessing.
• Your credibility is real when you use your own product. Every conversation with a user carried credibility because I wasn't advising from theory. I was a reseller talking to other resellers. That authenticity matters.

The Real Lesson

This isn't a story about a kid who cleverly funded his startup by reselling.

It's a story about constraints forcing better product decisions.

I couldn't afford servers and AI training. So I needed a revenue model. The only revenue model I understood was reselling. So I resold. And reselling with my own product meant I immediately knew which features mattered and which were nice-to-have.

That's backwards from how most products are built. Most founders start with theory: "Resellers probably want X." I started with necessity: "I need $2,000–5,000/month or my infrastructure dies. What will help me make that?"

Those are very different starting points. And they lead to very different products.

For Resellers Reading This

The reason Botifex doesn't feel like other reseller tools is that it was built by someone who had to use it to pay his infrastructure bills.

I know the pain points because I've lived them. Every feature exists because it directly impacted my ability to source deals, move inventory faster, and generate revenue.

When I recommend a feature, it's not marketing. It's "I used this to make $300 on a MacBook at 1 AM" or "This saved me 20 minutes per listing, which meant I could list three times as many items."

I don't theorize about what resellers need. I've been a reseller paying real costs.

Start free. See if it works for you. If it does, one flip typically covers the monthly cost.