Daniel Detwiller_

Full-stack Engineer · Boston, MA

3,737 contributions in the last year·GitHub

Scroll

About

I’m focused on building thoughtful, reliable products. I’ve worked across web apps, mobile apps, internal tools, data-heavy systems, and early-stage startups, and I enjoy fast-moving environments where I can balance clean user experiences, practical engineering, and getting the details right. I like getting deeply invested in what I’m building and figuring out how to make it genuinely good, not just functional.

I started coding at 15 because of Minecraft. I built mods and Java plugins, and ran a server that saw 250+ players at one time (at 16). That’s what got me into building software, and I haven’t stopped since.

Outside of work I like to surf, build, invest, and health max. You’ll always find me wearing a Whoop.

Experience

Athlix

Founding Engineer ·2025 – Present

An all-in-one platform for fitness coaches, spanning marketplace discovery, client management, billing, messaging, and mobile experiences. Built and shipped the full product across web, mobile, backend systems, and infrastructure. Four surfaces sharing a single API, data model, and design system, with an AI layer that coaches can talk to and agents that help run the company.

Web Marketplace

The public front door. Server-rendered coach profiles and search built for discovery, with onboarding and checkout that turn a visitor into a paying client in a single flow.

Athlix Coach Web App

Where coaches run their business: client roster, the drag-and-drop program builder, messaging, and billing. The heaviest surface in the product, and the one coaches live in every day.

Athlix App

The client-side mobile app. Assigned programs, workout logging and progress history, and a direct line to your coach. Designed to stay usable mid-set, one-handed, in a loud gym.

Athlix Coach App

The coaching companion on mobile. Check in on client activity, answer messages, and adjust programs away from a desk, with push notifications for the moments that need a fast reply.

Coach Agent

An assistant on web and mobile that can read and act on a coach’s workspace. Every capability declares who can use it, what plan it needs, and how risky it is. Anything that changes data comes back as a preview first, and confirming it runs exactly once through the same code and permission checks as the dashboard.

MCP Server

The same tools, opened up to Claude and ChatGPT. A coach signs in with OAuth and their own assistant can work with their clients, programs, and check-ins, scoped to exactly what Athlix would let them do. One implementation, not a second copy of the agent.

AI Workout Builder

A paid feature for clients training on their own. Describe the workout you want and Claude streams structured edits straight into the builder. Every edit is validated against a schema, salvaged when the model wraps it wrong, and capped per person so the cost stays predictable.

AI Workforce

A team of agents that lives in our Discord and helps run the company. A founding agent creates specialists, delegates work to them, and keeps long-term memory that tracks where every fact came from. They read from the product database, Stripe, GitHub, and Gmail, and anything that reaches outside the company waits for a human to approve it.

Alpenglow Maps

Founding Engineer ·2025 – Present

A framed ski trail map that hangs on your wall and lights up with the mountain’s live conditions. Every trail and lift gets its own LED, wired to the resort’s official report. Built end to end: the hardware and firmware, the conditions platform behind it, the design tool that turns artwork into a drill plan, the companion app, and the storefront that sells it.

The Map

An ESP32 driving addressable LEDs behind hand-built framed artwork. It joins home Wi-Fi and checks in every few minutes, no hub required. A master board owns the network, the clock, and the state, and replicates to follower boards that own the panels, so the display keeps animating while the network hiccups.

Conditions Platform

The service that watches the mountains: scrapers for ten New England resorts normalized into one schema, served through a deliberately tiny endpoint a memory-constrained microcontroller can afford to poll. Scheduling follows operating hours: a minute apart while the lifts are spinning, five while they’re not.

Self-Healing Scrapers

Resorts redesign their trail reports without warning. A break is captured with the raw payload that caused it and handed to a sandboxed coding agent in CI, which opens a pull request against the parser. It merges on its own only if it replays that payload, passes the test suite, and stays inside the files it’s allowed to touch. Anything else waits for me. Canaries recheck the live sites after every merge and every night. Bot challenges, tarpits, and resorts sharing one host each get their own handling.

Mapper

A CAD-lite editor for the physical build. Trace each trail over the artwork and it emits the firmware’s geometry header, a drill plan, and a drill mask, every position canonical in millimeters and every export checksummed. The order you click the LEDs is the order they light up.

Companion App

Setup and control on iOS and Android. It joins the map’s own hotspot to hand over Wi-Fi credentials, through a small native module I wrote because nothing off the shelf did it cleanly. After that it’s a remote: layers, brightness, effects, and a recorded ski day replayed as a comet running your tracks down the wall.

Hyperproof

Product Engineer ·2024 – Present

Compliance and audit software for frameworks like SOC 2, HIPAA, and ISO. I work across the stack, from product features to the services and CI underneath them, and I’m the main owner for customer-reported bugs and performance issues. I lead projects and mentor engineers on a team of four.

Faster CI

Built test impact analysis so CI only runs the tests a change can actually affect. It ran on a trial pipeline beside production before the cutover, and an audit of the next 1,766 builds showed median PR time down from 21 to 13 minutes, failures nearly halved, and not one test wrongly skipped.

Partner Sync

A .NET service on Azure Functions that keeps customer data in sync both ways with an audit partner’s platform, driven by Service Bus queues. I own it, and set up its deploys to swap slots so a release never takes it down.

Content Management

The pipeline that gets thousands of compliance frameworks and templates into every environment and keeps them current.

TALON

Senior Software Engineer ·2022 – 2024

A price transparency app that helped people with health plans find care and know what it would cost. I worked across the mobile app, the claims data behind it, and the tools that kept its prices honest, all under HIPAA, and led projects and mentored engineers on a team of four.

Mobile App Revamp

Rebuilt the consumer mobile app from scratch and owned it end to end. More than 2 million members used it.

Claims Pipelines

Python pipelines on AWS that worked through petabytes of payer claims and price transparency files, with HIPAA controls around the patient data inside them.

Provider Directory

A self-updating directory of more than 6 million providers that continuously pulled in sources like NPPES and HL7/FHIR feeds, with a field-level trust system that decided which source won when they disagreed. Provider data stayed accurate without anyone maintaining it by hand.

Price Backtesting

An algorithm and internal tool that checked the prices we pulled from payers’ published price files against what real claims actually paid. It turned quiet data bugs into something the team could see and fix, and brought price accuracy up to 95%.

ARUP Laboratories

Software Engineer · 2021

Prototyped patient-facing tools to help make medical charts and lab results easier to understand.

Stack

Education

University of Utah

B.S. Computer Science · 2022