Backend & APIs
Service boundaries, clean contracts, and maintainable application architecture.
Software engineer · San Diego, CA
Agentic and backend systems that stay reliable under pressure.
Agentic Engineering · Backend Systems · Production Reliability
With 6+ years in software engineering, I build resilient services and tool-using AI agents that plan, execute, verify outcomes, and recover when reality does not match the plan.
- Tool-using agents
- Production microservices
- Observability & recovery
system.topology
operational
request
response
CLIENT
Web / Mobile
EDGE
REST API
ASP.NET · Quarkus
ASYNC
Event Bus
idempotent flows
STATE
Postgres + Redis
INTELLIGENCE
Agent + RAG
tools · vectors · LLM
Profile
I work where application code meets production reality.
I’m a software engineer focused on agentic engineering, backend systems, and distributed architectures. At Alphatec Spine, I design and operate production microservices and APIs for a HIPAA-compliant medical cloud platform.
My day-to-day work includes eliminating race conditions, designing idempotent workflows, investigating failures across service boundaries, and improving API and database performance.
Outside of work, I ship SaaS products and real-time applications to explore agent planning, tool execution, RAG, MCP, browser automation, computer vision, WebSockets, and systems programming.
Technical capabilities
Tools organized by the problems they solve.
Production experience across the application lifecycle, from API contracts and data models to deployment, observability, and incident response.
Cloud & Distributed Systems
Reliable asynchronous workflows that tolerate retries, partial failures, and concurrency.
Agentic AI & Machine Learning
Tool-using agents that maintain context, verify execution, recover from failures, and stay grounded in source material.
Execution, Quality & Observability
Deterministic automation and measurable systems with fast feedback loops in development and production.
Selected work
Systems I’ve designed and shipped.
Each project explores a different systems problem: making autonomous work observable, grounding AI in context, turning conversations into structured data, and synchronizing real-time state.
01
Live demo
Agentic engineering · Browser execution
Tabvio
A watchable browser agent that plans and executes multi-step tasks, asks for human input when needed, and continues follow-up work in the same live browser session.
- Challenge
- Turn open-ended goals into reliable browser actions while keeping the agent grounded in what it can observe, verify, and safely execute.
- Architecture
- LangChain and Deep Agents run a LangGraph workflow with deterministic browser tools, human-in-the-loop interrupts, live execution events, and an observe-decide-act loop that rechecks page state before continuing.
- My role
- Designed the agent architecture, JavaScript page scanner, tool contracts, execution safeguards, conversation continuity, specialized navigation subagent, concurrent session lifecycle, and evaluation tests.
02
Live product
AI-assisted SaaS · Three-service architecture
Backlogr
Helps developers, QA engineers, and product owners refine software tickets before implementation by analyzing issues, comments, screenshots, and indexed source code.
- Challenge
- Surface missing requirements, ambiguous behavior, edge cases, and relevant implementation files before work begins.
- Architecture
- LLM agent workflows grounded by RAG across a Quarkus API, FastAPI AI service, Angular client, and Redis-backed coordination.
- My role
- Product design, distributed architecture, agent behavior, retrieval pipeline, full-stack implementation, and deployment.
03
Customer-facing SaaS
Conversational AI agent · Tool execution · MCP
NexMenus
Helps small food businesses create and update complete digital menus and storefronts through natural-language requests, reducing setup work for non-technical users.
- Challenge
- Turn incomplete conversational requests into reliable, multi-step menu changes without requiring users to understand the product’s backend operations.
- Agent design
- Built a custom Java agent loop with structured tool calling, validation, recoverable errors, and internal API execution. Exposed the same production tools through MCP so external AI clients can manage storefronts without duplicating application logic.
- Evaluation
- Built observability around real production conversations and turned recurring failures into test cases, improving task completion, reducing user friction, and making the agent more reliable over time.
04
Live multiplayer game
Real-time systems · 3D physics
Sumo Car
A multiplayer 3D driving game with physical steering, collisions, and a shared world synchronized in real time through a Rust backend.
- Challenge
- Keep independent browser clients synchronized while preserving responsive physics-based vehicle control.
- Architecture
- Three.js and Rapier run the 3D simulation while a Tokio-based Rust service manages sessions and broadcasts player state over WebSockets.
- My role
- Built the physics model, multiplayer protocol, concurrent server, shared-world state, and browser experience.
Additional work
Applied machine learning
Computer vision · Quality inspection
Surgical Screw Detection
Production object-detection pipeline using YOLOv8, PyTorch, Python, and OpenCV, with targeted retraining workflows for new failure cases.
Computer vision · Subject selection
FoodPhotoEnhancer
AI pipeline combining Grounding DINO and SAM to identify a dish, build a detailed mask, and select the most likely main subject for enhancement.
Experience
Building reliability into every layer.
Alphatec Spine
Software Engineer
- Design and develop production .NET microservices and REST APIs for a HIPAA-compliant medical cloud platform using ASP.NET Core, PostgreSQL, Redis, and Azure services.
- Use agentic engineering workflows to decompose production changes into scoped planning, implementation, review, and testing stages with human approval before merge.
- Redesigned a workflow spanning three backend services to eliminate race conditions using idempotent operations, resilient error handling, and explicit state transitions.
- Resolve distributed production incidents through Datadog and Application Insights, tracing asynchronous workflows, service interactions, and failure propagation.
- Improve API response times through query optimization, targeted indexes, and Redis caching.
- Contribute to code reviews and system design, and mentor engineers on backend design, debugging, and reliability practices.
Ace Parking
Software Engineer — DevOps & Automation
- Rebuilt unstable CI/CD infrastructure using Jenkins, Groovy, Docker, Ubuntu Linux, and AWS, improving deployment reliability and build stability across multiple products.
- Developed API integration workflows with JavaScript and Postman and scalable browser automation with Selenium, Java, JUnit, and Katalon.
Dexcom
Software Test Development Engineer I
- Tested BLE and Wi-Fi glucose-monitoring devices and companion mobile applications in a regulated medical-device environment.
- Developed iOS and Android automation frameworks using Java, JUnit, and Appium to expand regression coverage.
Foundation
Education & credentials
Computer Software Engineering
Java Programming · UC San Diego Extended Studies
Agentic & Cloud Engineering
LaunchCode · Agentic Engineer Training Program, in progress
LangChain Academy · LangChain and Deep Agents foundations
AWS · Cloud Technical Essentials
Verify credentialContact
Let’s build something that lasts.
If you’re working on backend platforms, distributed systems, or applied AI, I’d like to hear about it.
jimmy-galvan@live.com.mx
San Diego, California