José Manuel Olivar Herrera

Software engineer

José Manuel

Olivar Herrera

“I build the systems that have to stay up — five years on the platforms a national electoral process depends on: architecture, identity, data, and continuous delivery.”

Java · Spring Boot · Kubernetes · Azure · Keycloak · Python

Profile

Current
Software engineer · R&D · Byte, Guatemala
Languages
Java · Python · C# · JavaScript
Depth
Distributed services, identity, and delivery under availability pressure
Scale
National — electoral infrastructure, public–private delivery
Education
M.Sc. Computer Systems Security · UMG · 2024
Speaks
Spanish (native) · English (professional)
Availability
Guatemala — remote or hybrid

Built to stay up.

Five years on platforms where failure is public: microservices, Kubernetes on bare metal, hybrid Azure — engineered for the day everyone shows up at once.

Security by formation.

A master's in systems security behind every design decision — identity with Keycloak and JWT, IAM, and the habit of asking what breaks first.

Immovable deadlines.

Elections don't reschedule. I've shipped national-scale systems on dates that could not move, with CI/CD pipelines that cut the downtime windows.

Selected work

Public-sector platforms at national scale. Most of it lives behind institutional walls, so here is what it did rather than a link.

Pre-election preparation services

Took the preparation platform from a monolith to microservices — the flows that have to be correct and fast before a national process opens. More headroom to scale, better response times under load.

Spring Cloud Gateway · Eureka · Kubernetes · MetalLB

Tribunal Supremo Electoral 2020—2024

SIDI — nationwide media distribution

Ran the public–private seam: partnered with an external vendor to land SIDI and integrate it with TSE systems, keeping the media channel operable at national scale.

REST APIs · Messaging · SFTP · Vendor liaison

Tribunal Supremo Electoral 2020—2024

Citizen certification & documentation

A system for issuing citizen certifications and documents that cut request times and removed the manual work stacked around them.

Java · Spring Boot · Oracle · SQL Server

Tribunal Supremo Electoral 2020—2024

Deceased-citizen records, with RENAP

Automated deceased-citizen processing against the national registry — fewer manual errors, a cleaner electoral register.

Batch integration · Oracle · Stored procedures

TSE · RENAP 2020—2024

Experience

Byte

Software engineer · R&D · Guatemala

2025—present

  • Design and implement APIs and services in Java, Spring Boot, and Maven: cross-system integration, stable contracts, tests, and repeatable deploys.
  • Refactor legacy code — bottlenecks, technical debt, performance, maintainability — pairing with tech leads and QA through high-pressure cycles.
  • Fold AI agents and LLMs into the engineering workflow: implementation, review, documentation, system analysis — with tool calling and human oversight.
  • Collaborative Git — feature branches, merge and rebase, conflict resolution — and agile practice tuned for stable delivery.

Tribunal Supremo Electoral

Software development process manager · Guatemala

2020—2024

  • Redesigned pre-election preparation systems toward microservices with Spring Cloud Gateway and Eureka, lifting performance and availability in the flows that run before the process.
  • Acted as liaison between the public sector and a private vendor to land SIDI — nationwide media distribution — and integrate it with TSE platforms.
  • Brought Azure into the on-prem estate, with GBM, enabling hybrid scale-out for preparation services.
  • Implemented CI/CD with Azure DevOps and Azure Pipelines, shortening release cycles and cutting downtime windows.
  • Hardened access with JWT and Keycloak, and built REST APIs and messaging for day-to-day operation of the preparation systems.
  • Operated Oracle and SQL Server — stored procedures, query tuning — and critical file transfers over SFTP in Java.

Applied AI

Applied practice with language models and agents in software engineering: breaking work down, calling tools, iterating on code, and evaluating output before it ships.

Service prototypes that consume LLM APIs with versioned prompts, guardrails, and evaluation — grounded in a master's in systems security. The interesting part is not the generation; it is the boundary you put around it.

Agents
Orchestration, chained steps, bounded context, human review
Tools
Tool calling over Model Context Protocol
Daily
Claude Code, Codex, and Grok inside real delivery work
Limits
Security, cost, and latency judged before anything ships

Stack

Languages
Java · Python · C# · JavaScript
Platforms
Spring Boot · Maven · Spring Cloud Gateway · Eureka · REST APIs · Microservices · Express.js · React
Data
Oracle · SQL Server · Redis · pandas · NumPy · scikit-learn · Power BI
Cloud & infrastructure
Kubernetes (bare metal) · MetalLB · Azure · AWS · Databricks · Azure DevOps · Cloudflare · Prometheus · Grafana · SFTP
Security
JWT · Keycloak · IAM · High availability
AI
Agents · LLMs · Tool calling · Orchestration · Prompts & evaluation
Practice
CI/CD · Agile & Scrum · Git (merge, rebase) · Public–private liaison

Education & credentials

M.Sc. Computer Systems Security

Universidad Mariano Gálvez · completed 2024

B.S. Information Systems & Computer Science

Universidad Mariano Gálvez · Ingeniero

Cisco Data Science Essentials with Python 2026
Cisco Introduction to Modern AI 2026
Cisco Introduction to Data Science 2026
INTECAP Statistics for Data Analysis 2026
INTECAP Power BI Intermediate 2020—26
INTECAP Artificial Intelligence and Data Science 2020—26

Contact

hello@joseherrera.dev
GitHub github.com/manuolivarh LinkedIn linkedin.com/in/josemaolivar