Full Stack React Software Engineer (AI Infrastructure) Role in St. Petersburg, FL

Noblesoft Solutions, Inc.; Saint Petersburg (FL); On-site; USD 120,000 - 180,000; Full time

Job summary

Noblesoft Solutions, Inc. in Saint Petersburg, FL, seeks an experienced software engineer to own end-to-end features across the Nash platform, from React/TypeScript frontend through Java/Spring Boot services to the Oracle schema behind them.

You will extend the asynchronous LLM execution pipeline, integrate Azure OpenAI and AWS Bedrock, and implement rate limiting, token accounting and observability using Splunk, Datadog and Dynatrace.

Qualifications

• 5+ years building and running production web applications end to end.
• React 18 + TypeScript with experience in high-volume data grids.
• Strong SQL and relational DB skills including schema design and migrations.
• Solid grasp of asynchronous and concurrent programming concepts.
• Git-based workflows with CI/CD pipelines.
• Ability to write for a non-engineering audience during change control.

Responsibilities

• Build features end to end across the Nash platform from frontend to Oracle backend.
• Extend the asynchronous LLM execution pipeline with retry and error handling.
• Integrate model providers across Azure OpenAI and AWS Bedrock and report usage.
• Design throughput controls to scale toward 1M+ requests/day.
• Own usage, cost and token accounting; surface metrics in admin UI.
• Deliver self-service authoring for business teams.
• Manage changes with governance gates and controlled promotions.
• Build observability with logging and dashboards (Splunk, Datadog, Dynatrace).
• Harden the platform with auth, secrets handling, and dependency currency.
• Collaborate with DBAs, platform, networking and compliance teams.

Skills

Production web apps

React 18 + TS

SQL & Relational DBs

Async/concurrent programming

Git CI/CD

Communication for non-engineers

Tools

Kubernetes

Docker

Oracle PL/SQL

Azure OpenAI

OpenAI API

IBM MQ / Kafka

Job description

• Build features end to end across the Nash platform — from the React/TypeScript front end through the Java/Spring Boot services to the Oracle schema behind them.
• Extend the asynchronous LLM execution pipeline. Work on the message-driven services that accept requests, queue them by priority, call the model provider, and return results — including retry, failover and error handling semantics.
• Integrate and maintain model provider connections across Azure OpenAI and AWS Bedrock: new model onboarding, routing, provider SDK upgrades, and the differences in how each provider reports usage and errors.
• Design and implement throughput controls — token- and request-per-minute rate limiting, backpressure, and circuit breaking — so the platform scales toward its 1M+ requests/day target without exhausting provider quota.
• Own usage, cost and token accounting. Extend the data model and reporting that tracks input, output and reasoning tokens per request, and surface it in the admin UI.
• Deliver the self-service authoring experience that lets business teams create and tune AI use cases without hands-on engineering support.
• Work the full change lifecycle in a regulated environment — design, code review, governance and compliance gates, and controlled promotion across environments.
• Build and maintain observability — structured logging, service metrics, and dashboards in Splunk, Datadog and Dynatrace; diagnose production issues across a distributed, queue-based system.
• Harden the platform — service authentication and authorisation, secrets handling, and dependency currency.

Collaborate with DBAs, platform, networking and compliance teams, whose sign-off the work depends on.

Skills:

**Required**

• 5+ years building and running production web applications end to end
• **React 18 + TypeScript** — hooks, modern state management (Redux Toolkit / RTK Query), and experience with a high-volume data grid (ag-grid or equivalent)
• Strong **relational database and SQL** skills — schema design, query tuning, and working with DBA-applied migrations rather than ORM-generated DDL
• Solid grasp of **asynchronous and concurrent programming** — thread pools, executors, and the failure modes that come with them
• Git-based workflow with peer review; CI/CD pipelines
• Able to work within formal change control, and to write for an audience that includes non-engineers

**Strongly preferred**

• **Project Reactor / reactive Spring** (`Mono`, `Flux`, WebFlux) — core to the execution services
• **Message-oriented middleware** — IBM MQ, JMS, ActiveMQ or Kafka; consumer concurrency, redelivery, dead-letter handling
• **LLM API integration experience** — Azure OpenAI, Bedrock or OpenAI directly; prompt handling, token accounting, rate limits, and 429/backoff semantics
• **Kubernetes and Docker** in an enterprise setting
• **Micro-frontend architecture** — Webpack Module Federation or similar
• Oracle specifically, including PL/SQL
• Observability tooling: Splunk, Datadog, Dynatrace, Prometheus/Micrometer

**Nice to have**

• Azure API Management, or API gateway policy work generally
• Financial services or another regulated industry
• Performance and load testing at scale
• Maven and private artifact repositories (Artifactory)