Skip to content
Currently building

Kyle Schneiderengineer buildingproducts that

I design, build, and ship production apps end-to-end — from Claude Vision pipelines that turn photos into 3D-printable house models to FAA airspace screening and crew-dispatch software for fence contractors. Here's a collection of live AI, geospatial, and vertical SaaS products, all on one modern stack.

See the workLinkedInGitHubEmail
Kyle presenting at Card Tech+
Now building
01 / 07
0
Projects shipped
0+
Years building
0
Stacks fluent in
∞
Coffees today
Personal projects — 2024 — 2026

Things I've beenbuilding lately.

Each one solved a real problem for a real audience — from aviation engineers to fence-company owners to homeowners who want a 3D-printed model of their house on the mantel.

Live preview
Your House in 3D live homepage
Live

Your House in 3D

Turn a handful of phone photos into a watertight, 3D-printable model of your home.

Upload a few photos of your house and get back a printable STL — no scan rig, no CAD. Claude Vision runs a multi-pass elevation analysis to read the geometry straight from the images, then a Python FastAPI service builds a watertight mesh with CSG (manifold3d / trimesh) and validates it for printing. A Next.js storefront with Stripe checkout drives the order; Google Maps handles address autocomplete only — the geometry comes entirely from the photos.

  • Multi-pass Claude Vision elevation analysis derives geometry from photos alone
  • CSG meshing (manifold3d / trimesh) produces watertight, printable STL files
  • Print-readiness validation before a model ships to the slicer
  • Stripe checkout storefront with Google Maps address autocomplete
Next.js 16Claude VisionFastAPImanifold3dtrimeshDrizzle ORMAurora Serverless v2StripeSST 4
Live preview
FenceFlow live homepage
In progress

FenceFlow

The operating system for mid-market fence companies.

FenceFlow replaces the fragmented ArcSite + FenceCloud + QuickBooks + Google Sheets stack with a single workflow-aware platform. Estimators draw on a map, generate signed estimates in under 15 minutes, and the owner sees weekly profitability at a glance.

  • Draw a fence line on a map → a geometry engine prices it in three modes
  • Full RBAC: Owner, Sales, Foreman, and Customer portals
  • Real-time scheduling and crew dispatch
  • Comprehensive Vitest + Playwright test suite
Next.js 16React 19Drizzle ORMAurora Serverless v2Auth.js v5CognitoSST 4Playwright
Live preview
LOS Analyzer live homepage
Shipped

LOS Analyzer

Screen a proposed structure for FAA Part 77 penetration and control-tower line-of-sight in minutes.

Pick an airport, drop in a proposed structure, and LOS Analyzer tells you whether it penetrates the FAA Part 77 imaginary surfaces or breaks the control tower's line of sight — then generates a PDF report and a draft FAA Form 7460. A terrain-aware 3D engine models earth curvature and atmospheric refraction so the sightline math matches reality, validated against named FAA cases. Built on a Google Maps satellite overlay with OCR-assisted coordinate intake.

  • FAA Part 77 surface-penetration screening for any proposed structure
  • Terrain-aware 3D line-of-sight with earth-curvature + atmospheric refraction
  • Auto-generates a PDF report and a draft FAA Form 7460
  • Validated against named FAA cases; $499/report via Stripe
Next.js 16Express 5AWS LambdaAstroDrizzle ORMAurora Serverless v2Turf.jstesseract.jsStripeSST 4
Live preview
Sygmatic live homepage
Live

Sygmatic

Authentic language learning from real YouTube videos.

Sygmatic teaches languages from real native-speaker YouTube clips — every word transcribed, every sound visualized as a waveform, every lesson followed by a quiz auto-generated from the speech's grammar and tense. Originally a paid SaaS at sygmaticlanguage.com; modernized in 2026 as a portfolio piece on Next.js 16, Drizzle, Auth.js v5, and SST 4.

  • 100+ authentic video lessons across English, Spanish, German, French
  • Word-level transcript with click-to-seek and per-word definitions
  • Waveform peaks pre-generated from YouTube audio and committed to git
  • Quizzes auto-generated from each lesson's part-of-speech and tense metadata
Next.js 16React 19Drizzle ORMAuth.js v5Wavesurfer v7Aurora Serverless v2SST 4
Live preview
Landscape AI live homepage
In progress

Landscape AI

Snap a photo of your yard, pick a style, and see a photorealistic redesign of your real yard.

A DIY tool for homeowners, not pros: upload a photo of your actual yard, choose a style, and a two-stage Gemini pipeline returns a photorealistic redesign that keeps your real yard intact and changes only the landscaping. Usage is token-metered with Stripe packs, on a production Next.js + Aurora stack.

  • Two-stage Gemini pipeline preserves the real yard, restyles only the landscaping
  • Photorealistic redesigns from a single homeowner photo
  • Token-metered usage billed through Stripe packs
  • Auth.js v5 + Cognito on an Aurora Serverless v2 backend
Next.js 16Vercel AI SDKGeminiDrizzle ORMAurora Serverless v2Auth.js v5CognitoStripeSST 4
Engineering standard

Every build runs onthe same backbone.

The projects above aren't one-off demos. Each one ships on the same modern foundation — so they're not just live, they're maintainable and observable.

State-of-the-art CI/CD

Push to main → GitHub Actions deploys automatically over AWS OIDC. No manual steps, no static keys.

Environment split

Isolated dev and prod stages, promoted independently so changes are tested before they ship.

Multi-language support

Internationalized from day one — copy and formatting ready for a global audience.

Light & dark mode

Theme-aware UI that respects the visitor's system preference, end to end.

All on AWS

Serverless on AWS — Lambda, CloudFront, S3, and Route 53 via SST / OpenNext.

Ops dashboard

Traffic, errors, and status — all in one place.

Every deploy reports into a single ops board that auto-discovers each project in AWS. One screen shows what's healthy and what needs attention — no jumping between consoles.

Traffic

Lambda invocations / 24h

Errors

Error rate + p50 latency

Status

Route 53 uptime checks

Deploys

Last-deploy SHA + cost

Contracting — client work

Built for clients,on contract.

Paid engagements where I parachute into someone else's codebase or greenfield a product end to end — from AI SaaS for government contractors to firmware for battery-powered IoT sensors.

Live preview
BreezeRFP live homepage
Shipped

BreezeRFP

AI platform that helps companies win government contracts.

A contracting build (shipped as JustWin): an end-to-end SaaS for government-contracting teams. A Celery pipeline pulls solicitations from the SAM.gov API and an LLM matching algorithm ranks them by winnability. To respond, a LangChain + FAISS retrieval pipeline grounds an AI assistant (Claude 3.5 Sonnet and GPT-4o) in each company's knowledge base, drafting proposal outlines, cover letters, and FOIA requests straight from a lead's RFP documents. Vue 3 SPA on a Django + DRF backend.

  • Celery workers ingest solicitations from the SAM.gov API and rank them by winnability with an LLM matching algorithm
  • LangChain + FAISS RAG over each company's knowledge base grounds the AI assistant in their own past work
  • Claude 3.5 Sonnet and GPT-4o draft proposal outlines, cover letters, and FOIA requests, plus a smart reader that distills long RFP PDFs
  • Django + DRF API with PostgreSQL, Redis, and Stripe billing — deployed on Heroku with document storage on AWS S3
Vue 3VuexDjangoDRFPostgreSQLRedisCeleryLangChainFAISSClaude 3.5 SonnetGPT-4oAWS S3HerokuStripe
Live preview
Smart Plank live homepage
In progress

Smart Plank

Embedded firmware development for connected hardware.

An ongoing contracting engagement (under NDA) doing embedded firmware development for connected hardware devices. The product details stay private, but the work is hands-on firmware engineering — microcontroller bring-up, on-device sensors, wireless connectivity, and over-the-air updates, taking a device from bare metal to talking to the cloud.

  • Microcontroller firmware bring-up in embedded C/C++
  • On-device sensors, wireless connectivity, and over-the-air updates
  • Hands-on hardware debugging with modern embedded toolchains
Embedded CC++FirmwareMicrocontrollersBLEWi-FiOTA
Day job — Capital One

What I shipat the bank.

My day job is engineering at Capital One — high-throughput backend services, conversational voice platforms, and partner-facing card experiences at bank scale. A few things I've worked on:

Shop with Rewards

External APIs — Go, up to 250 TPS

On the dev team that spun up the external Go APIs powering Shop with Rewards, letting cardholders redeem rewards directly at checkout. Engineered for up to 250 TPS with the fastest response times we could squeeze out.

That's my manager, Sridhar Rajaram, featured in the video.

Omilia Voice IVR

Conversational IVR platform

Helped move Capital One's IVR from the legacy 24/7 vendor over to Omilia — building OL and Flows in Omilia for callers, and supporting the platform that lets Capital One deploy changes onto Omilia.

~1Mcalls to Capital One a day
Read about Omilia

Partner card sites

Co-brand cardholder experiences

Built menards.capitalone.com and other partner co-brand card experiences — the cardholder-facing sites for Capital One's retail partnerships.

Visit the site
Physical projectsMore coming soon

3D-printed lighting,
designed and built.

Software pays the bills, but I spend evenings designing custom lighting fixtures in CAD, dialing in print profiles, and wiring them up into finished pieces. Each one is a full pipeline from sketch to switch — and it's the first of more physical builds on the way.

Read the writeup@3d_printed_lighting
About

I like buildingunsexy softwarethat pays the bills.

Every project on this page started one of two ways: a friend or family member came to me with an idea they couldn't get built, or I ran into something in my own everyday life that I figured deserved a dedicated solution. None of them began as a business plan — each one started as a real, specific need from someone I know, myself included.

From there the playbook is always the same: learn the workflow cold, then build it for real — Next.js and TypeScript on top, Postgres for state, and Python for the heavy lifting (3D mesh generation, vision pipelines). I care more about clean state machines and a tight feedback loop than I do about chasing frameworks.

Off-hours: designing and printing 3D lighting fixtures, tinkering with side projects, and always sketching the next one.

Kyle pitching Sygmatic at a Madison startup competition

Ship-first mindset

Four products in flight covering AI, geospatial, vertical SaaS, and consumer hardware. I move fast and finish.

Full-stack range

Comfortable from Next.js Server Components down to FastAPI mesh-generation services and Stripe webhook plumbing.

AI-native

Building production pipelines with Claude Vision, structured output, and async job systems. Not bolt-on chatbots.

Built from real needs

Every project traces back to a friend, a family member, or a problem I hit myself — never a spec dreamed up in a vacuum.

In between projects

Off-screen.

Presenting Sygmatic, a language-learning startup
Sygmatic pitching at a Madison startup competition
Sygmatic team with competition awards
Presenting Sygmatic, a language-learning startup
Sygmatic pitching at a Madison startup competition
Sygmatic team with competition awards
Presenting Sygmatic, a language-learning startup
Sygmatic pitching at a Madison startup competition
Sygmatic team with competition awards
Always up for a good conversation

Got a hard problemworth building?

Questions about how I built something, thoughts on an approach, or just want to chat about unsexy software that pays the bills — always happy to connect.

Connect on LinkedInSee more on GitHub