About
Samuel Hernández
I'm a software engineer focused on backend systems, domain-heavy applications and software where correctness matters.
I am extending this work toward AI systems where evaluation, observability and controlled uncertainty matter.
How I got here
I started building full-stack applications: a geospatial web application in a university research group, then a document management system for the oil and energy sector and internal business applications for an industrial services company.
Over time the interesting part moved from the screens to what sits underneath them: access control for confidential documents, architecture decisions written down as records, and written comparisons such as GraphQL versus REST for specific services.
My current work is an internal engineering calculation and configuration platform. Engineering rules validated by specialists run in one Python and FastAPI backend, kept separate from changing operational data and from the commercial system.
How I work
A few habits that show up in most of what I build.
One authoritative place for each rule
Rules that matter live behind one implementation and one validation boundary. The interface checks inputs for usability; it does not keep its own copy of the rules.
Separate what is deterministic from what changes
A calculation from inputs and rules should be reproducible. Recommendations that depend on catalogs, stock or other live data are a different kind of answer, and I keep them apart.
Make system boundaries explicit
When several systems touch the same workflow, each one should own a clear part of it, and the data that crosses between them should be deliberate.
Domain specialists own correctness
I implement rules; the people who own them validate them. Tests built from their reference cases matter more to me than a coverage percentage.
Say what is not known
When something is not measured or not decided, I say so instead of covering it with architecture vocabulary.
Experience
Roles, dates and what I built in each.
Cables del Golfo MX
Software Developer
Domain discovery with engineering, commercial and operations stakeholders, and the design and implementation of an internal engineering calculation and configuration platform.
- Python/FastAPI backend with a deterministic calculation engine and server-side validation, implementing engineering rules validated by domain specialists and derived from published lifting-equipment standards.
- Validation of technical configurations and component compatibility using current availability data.
- React/TypeScript configuration workflow and PostgreSQL persistence for saved technical configurations.
- Docker-based setup for development and deployment.
Heavy-Lift Rigging Services & Consulting Mexico
IT Consultant & Software Developer
IT engagements from requirements to delivery: workshops, vendor evaluation and software architecture.
- Architecture decision records (ADRs) and solution architecture reports.
- A written evaluation of GraphQL vs. REST for ASP.NET Core microservices, comparing performance and scalability.
- An event-driven architecture design for a hybrid affiliate platform.
- Post-implementation review templates.
Software Developer
- Internal web applications with ASP.NET Core (C#), Angular and SQL Server.
- Power BI dashboards with KPI reports for the finance and operations teams.
- Angular and Tailwind CSS interfaces using RxJS.
- Flutter (BLoC) mobile inspection app for field data capture, with synchronization across sites.
GUMEC Soluciones Tecnológicas
Software Developer
Full development cycle of a document management system for the oil and energy sector.
- Document management system (Laravel, Vue.js, MySQL) that moved legacy processes to a platform hosted on IONOS Cloud.
- Data-processing modules for oil-well performance analysis, with encryption and role-based access control.
- VBA tools that check compliance with the NOM-024-STPS-2001 occupational safety standard.
- Production deployments and environment monitoring on cloud infrastructure.
Tecnológico Nacional de México — Instituto Tecnológico de Villahermosa
Programmer / Research Developer
- Geospatial web application with PostgreSQL, Django and React (Remix).
- Augmented-reality prototype in Unity/C# for HoloLens.
- Led InnovaTec, a real-time sign-language translation prototype.
Current direction
Current work includes studying and building around retrieval, structured outputs, agents and evaluation. The part that interests me most is the engineering around probabilistic components: checks that stay deterministic, evaluation against reference answers, and visibility into what a system accepted and why.
That is a direction, not a finished transition. I'm building it on the same backend, data and correctness work described above.
I'm interested in roles where software engineering and AI systems meet: backend and software engineering, implementation and forward-deployed engineering, and applied AI engineering as the direction of that work.
Independent client work goes through BKODE, separately from this site. bkode.lat (opens in a new tab)
Contact
Open to backend and software engineering roles, including work on applied AI systems. Remote.
sam.hernandezc103@