Welcome to My Portfolio 👋
¡Buenas! I'm Fernando Soto—a Full-Stack Software Engineer based in Chicago with a B.S. in Computer Science & Mathematics from Dominican University, rooted in my Mexican heritage.
I build efficient, scalable solutions across C#, React, SQL, Python, and more—from web apps to automation tools, always focused on impactful results.
Before 2024, I had zero athletic background. That year became something else entirely—a simple cycling century quickly spiraled into pushing limits I didn't know I had.
I tend to go further than planned. A century became 400 miles. An Everest became a ROAM. A first ultra-endurance race became a chase for 2nd place. In a world where finishing feels like enough, I keep asking what's next—and that's what I bring to engineering.
Accomplishments
In the short time I've been in the world of ultra-endurance, I've accomplished a lot. Below is a hand-picked selection of my proudest moments!
Spartan Trifecta
spartanA friend introduced me to Spartan races and offered to help me learn the obstacles. I signed up for the Trifecta—all three races in one weekend—before practicing a single one. He couldn't make it. I went anyway. Two days, three races, every obstacle cold. My feet had never experienced anything like it—but I finished. Five medals in one weekend. Five times my entire cycling collection. 😅
Mishigami 2025
My first ever race—and my first time seeing a real hill. Fought from the back of the pack all the way to 2nd place, on 2 hours of sleep, in a city that had never once prepared me for actual elevation.
After speedrunning the Strava run milestones, only one challenge remained: an ultramarathon. To mark it, I crafted a GPS run art route as a birthday tribute to a friend ahead of one of his toughest Spartan Races.
Marathon
runningSpeed-ran Strava's cycling milestones, then tackled the two remaining: a marathon and an ultra. Not a runner—but proved it anyway.
Everest ROAM
cyclingAn Everest—but harder. More elevation, more distance, stricter time limit. Over 1,150 hill repeats on the same bridge I completed my first Everest!
24-hr quad-state ride
cyclingBuilt up distance across the year: 100 → 150 → 200 → 250 miles, then pushed to the self-supported upper limit. Crossed Michigan, Illinois, Indiana, and Wisconsin in 24 hours.
Climbed the elevation equivalent of Mount Everest on a local Chicago hill (bridge 😅)—no sleep between attempts.
First American Century
cyclingAfter a bicycle accident that left me with a broken wrist, broken thumbs, and chipped teeth, I came back with one goal: 100 miles. Built up 30 → 40 → 60 mi, skipped 80, and ripped the bandaid.
What's Next
Skills
Experience
- Optimized high-throughput API workflows through profiling, SQL tuning, and business-logic refactors, improving performance by 7.5x and reducing query volume by 8x while supporting 3,000 req/s.
- Designed and enhanced bulk Time Entry and Pay Type APIs across versioned endpoints, supporting complex modifiers, duration-based workflows, and validation-heavy edge cases.
- Identified a misconfiguration gap that would have broken millions of users before shipping—flagged the need to query affected users first, preventing a high-impact regression.
- Resolved critical production and release-blocking issues including a cost center validation bug praised by management, audit trail gaps, payroll modifier logic, and logging failures—improving reliability for downstream consumers.
- Improved API usability through error formatting improvements on bulk endpoints, stronger validation logic, refined error messaging, and expanded Swagger documentation—reducing support overhead from partner teams.
- Increased platform reliability by migrating CI pipelines to dedicated build pools and completing .NET upgrades for internal services.
- Built and expanded automated test coverage for a core product suite used by ~22.5% of the user base, improving regression confidence and release stability.
- Migrated end-to-end testing from Selenium to Cypress (TypeScript), reducing test flakiness and accelerating test development velocity.
- Established a unit test pattern for framework middleware, creating a repeatable standard adopted by the team and added to the definition of done for new and updated functionality.
- Identified and resolved a browser-specific defect impacting 18% of users, restoring functionality for international customers.
- Created the TimeEntriesV2 OPTIONS endpoint and documented all restriction codes in Swagger, making a complex endpoint significantly easier to consume and reducing code review overhead.
- Proposed and drove adoption of a repeatable API documentation pattern—a filter table clarifying field definitions and operations—standardizing how the team documents endpoints.
- Implemented application tracking using Google Analytics to enable data-driven insights into user behavior and feature usage.
- Diagnosed and resolved two critical performance-related bugs, improving page load times by 15%.
- Collaborated with senior engineers to analyze stored procedure usage and support a broader database deprecation and modernization effort.
Projects
A fully offline React/TypeScript planning tool and Dockerized FastAPI service for modeling multi-day ultra-endurance cycling strategies—GPX analysis, terrain metrics, pacing, and map rendering all run in-browser with no backend required. Built to plan Mishigami 2025, where I finished 2nd place as the first Illinoisan to finish sub-four days.
- Engineered a GPX processing pipeline that decimates tracks of 100k+ points to a configurable render resolution, then dynamically re-samples at higher fidelity on zoom—avoiding aliasing artifacts at high magnification while keeping the DOM lean enough to handle full-course data without a backend.
- Computed per-split terrain metrics (elevation gain/loss, average grade, steep distance %) directly from raw GPS coordinates—data used in real race planning that contributed to a 2nd-place finish at the Mishigami Challenge, the first sub-four-day finish by an Illinoisan.
- Built a fully offline-capable app with full form state persisted to localStorage and raw GPX binary to IndexedDB, keyed by filename. Cross-session hydration merges saved states with current defaults rather than rejecting them, keeping the app forward-compatible as features evolve.
- Designed the pacing engine as a standalone, independently testable
package with its own DTOs and models—then wrapped it in a
FastAPI service with a single POST endpoint, making the same logic
available as both an in-browser port and a server-side fallback
selectable via
?engine=api. - Integrated OpenStreetMap's Overpass and Nominatim APIs—no key
required—to query nearby amenities and resolve GPS coordinates
to city labels, feeding a configurable Auto-Name system that
batch-generates human-readable segment names from templates like
{start} → {end}. Requests are debounced to respect rate limits without any server intermediary. - Built a React context signal pattern using a
{ segIdx, splitIdx, rev }object with a monotonicrevfield to drive cross-component navigation—guaranteeing re-fire even when the target index is unchanged—powering “Go to split” map pin interactions, auto-paging pagination, and scroll-to-expand form behavior from a single state shape.
A full-stack financial system that evolved from a database-driven tool into an API-first architecture with a modern web client, built for real-world daily use.
- Engineered a full-stack application, transitioning from a Windows Forms client to a decoupled React frontend backed by a RESTful API.
- Designed a normalized SQL database with stored procedures and triggers to enforce data integrity and automate financial transaction workflows.
- Developed a RESTful API in C# using service layer and DAO patterns, enabling scalable and maintainable business logic.
- Implemented a responsive React UI with pagination, validation, and error handling to improve usability and accessibility.
- Continuously iterate on the system through real-world usage, shaping features around practical lending scenarios and financial tracking needs.
Built during the GPU shortage to track Best Buy inventory and automate purchase readiness, navigating real-world bot detection and high-demand stock drops.
- Developed a Python Tkinter application to monitor item stock in near real-time with a lightweight, configurable desktop interface.
- Implemented web scraping logic resilient to bot-detection mechanisms, enabling reliable tracking during high-traffic drops.
- Designed a flexible polling system with configurable intervals and launch modes for continuous or targeted monitoring.
- Integrated notification systems ranging from audio alerts to push notifications to enable rapid response to stock availability.
- Successfully used the tool to secure a GPU during a live Best Buy drop, validating real-world effectiveness under constrained conditions.
