AI Software Engineer (Full-Stack + LLM)

About Beyond Key:

We are a Microsoft Solution Partner and a Great Place to Work-certified company. "Happy Team Members, Happy Clients" is a principle we hold dear. We are an international IT consulting and software services firm committed to providing. Cutting-edge services and products that satisfy our clients' global needs. Our company was established in 2005, and since then we've expanded our team by including more than 350+ Talented skilled software professionals. Our clients come from the United States, Canada, Europe, Australia, the Middle East, and India, and we create and design IT solutions for them. If you need any more details, you can get them at https://www.beyondkey.com/

About the role:

Need a software engineer who builds full-stack product features in Node.js, Python, and Next.js and ships the AI capabilities inside them. You will own features end to end, from APIs and UI to the LLM calls, prompts, and evaluations behind them. You will work directly with the engineering team and report to the engineering lead.

This is a hands-on individual contributor role. Development is the core of the job; prompt management is part of it. The right person writes clean production code and treats prompts the same way: versioned, tested, reviewed, and owned.

Responsibilities:

• Build and ship features across the dashboard: backend services in Node.js and Python, frontend in Next.js and React.
• Design and maintain APIs, data models, and integrations for a multi-tenant, multi-role application.
• Build AI features on LLM APIs (Anthropic, OpenAI, or similar): tool calling, structured output, RAG, and agentic workflows.
• Design dynamic prompting that assembles context at request time: role, permissions, preferences, history, and tenant settings, without hard-coding variants.
• Build and maintain prompt management in the codebase: a versioned prompt registry, templating, per-tenant and per-role variants, and rollback.
• Stand up evaluation and regression checks for AI features, so prompt and model changes ship with a measured result.
• Track cost, latency, and quality per feature and per model, and act when the numbers move.
• Own AI safety basics in your features: prompt-injection resistance, PII handling, and policy constraints across tenants.
• Write tests, review code, and keep CI/CD, observability, and documentation in good shape.

Required qualifications:

• 3+ years of professional software development, shipping production features end to end.
• Strong Node.js and TypeScript; solid Python for services, scripts, and AI tooling.
• Production experience with Next.js and React.
• 1+ year building with LLM APIs (Anthropic, OpenAI, or similar) in a shipped product, not only prototypes.
• Hands-on prompt work: dynamic prompt construction, templating, and runtime context injection, with prompts kept in version control.
• Experience testing AI output: golden sets, LLM-as-judge, or regression checks. Comfortable with REST or GraphQL APIs, SQL and NoSQL databases, and Git-based review workflows.
• Clear written communication: design notes, PR descriptions, and docs other engineers can maintain.

Nice to have:

• Experience with multi-tenant SaaS, especially per-customer configuration and role-based access.
• Background in RAG pipelines, vector stores, or agent frameworks.
• Familiarity with prompt-management or LLM observability tooling (for example Langraph, Langfuse, Braintrust, PromptLayer, Helicone, or an in-house equivalent).
• Exposure to AWS (Lambda, Bedrock, AppSync) or a comparable cloud stack.
• Real-time features (WebSockets, pub/sub) and event-driven architecture.
• Prior work on prompt-injection defense, red-teaming, or content policy enforcement.
• Experience with fine-tuning or distillation, and knowing when a prompt or code change is the better lever.