I build back-end platforms from scratch.
I'm Nathan — a Staff Software Engineer with a decade of production engineering across Rust, Ruby, TypeScript, Python, and Java. Today I'm the founder and principal engineer behind Skeptik, a portfolio of independent back-end products I designed, built, and deployed solo. Before Skeptik, I was the engineer teams handed the load-bearing problem to: the billing pipeline nobody could finish, the migration that had been deferred for a year, the integration that became core revenue.
Tier 1 to Staff, by way of finishing.
My formal engineering career started in 2018, but I'd been writing code on my own for two decades before that. The three years I spent in technical operations at Constant Contact — Tier 1 support up through Senior VOC Technical Engineer — are the part of my background I value most.
Customer support is where you learn what production actually feels like. Every escalation is a system telling on itself. Every irreproducible bug is a hidden assumption. By the time I switched into engineering, I'd already developed the instincts most engineers spend years acquiring: edge cases, failure modes, the gap between what a system claims and what it does.
The progression since then — Associate, Senior, Staff, founder — has been a single arc: take the hardest problem in the room and finish it. The systems I build are designed to outlast the team that shipped them.
A decade in production.
- 2015 – 2018
Customer Engagement → Senior VOC Technical Engineer · Constant Contact
Started at the bottom of the support stack and stayed long enough to learn what production actually feels like — every escalation, every bad deploy, every customer who couldn't tell us what was broken. Built internal tooling that lifted team throughput 15–20% and used it as the proof that earned an engineering seat.
- 2018 – 2021
Associate → Senior Software Engineer · Constant Contact
Sole developer of the Facebook Ads integration — a revenue-critical surface that I owned end-to-end. Co-led the company-wide React frontend migration and helped define the patterns adopted across teams.
- 2021 – 2022
Senior Software Engineer · Globalization Partners
Built the credits pipeline in the billing flow that the team had been unable to crack. Cut critical Postgres reporting query times by ~99% with systematic optimization. Automated payment reconciliation saved finance 10–15 hours weekly.
- 2022 – 2025
Senior → Staff Software Engineer · LinkSquares
Integration architect for the company's entire third-party offerings. Designed the event-driven (RabbitMQ) architecture that defined the org's transition to microservices. Cut deployment cycles from weekly hours-long monolith releases to on-demand 15–30 minute deploys. Mentored a team of 6.
- 2025 – 2026
Senior Software Engineer · LILT
Designed and deployed a worker-based ingestion pipeline for 50+ content connectors, replacing a fragile synchronous flow. Optimized the core job runner from 7s to 0.3s per file (~96% improvement). Operated as a generalist across Python, Java, TypeScript, and C#.
- 2026 – Present
Founder & Principal Engineer · Skeptik.io
Self-funded, solo-built portfolio of production back-end platforms spanning security, realtime, metering, and matching — deployed multi-tenant with codegen-driven SDKs across six languages.
Instincts I lean on.
I design the wire, the schema, and the failure modes before writing the application code. SDKs across six languages are codegen targets, not maintenance burdens.
Servers shouldn't see plaintext they don't need. ShrouDB and Herald both push E2EE to the client via WASM and treat sensitive bytes as opaque on the wire.
I'd rather ship a few sharp tools than one general-purpose suite. Adopting one of my platforms doesn't commit you to the others; the boundaries are deliberate.
You shouldn't discover a limit in production. Every system I build is documented for what it deliberately doesn't do, not just what it does.
What I'm building right now.
Production back-end platforms under Skeptik. Each solves one problem end-to-end with no suite lock-in.