Senior Full-Stack AI & Data Engineer (Java / SaaS)

Macrosoft Pakistan is looking for a Senior Full-Stack AI & Data Engineer (Java / SaaS)

Experience - 8+ years

Location - Remote

About the Role: We are looking for a versatile builder to own end-to-end features across our multi-tenant B2B SaaS platform. You will bridge enterprise Java backend engineering, real-time data pipelines, and modern AI orchestration to deliver reliable, production-grade AI workflows to our customers.

Key Responsibilities:

AI & Backend Engineering:

- Architect and build scalable microservices in Java 21+ and Spring Boot, integrating LLMs using Spring AI and LangChain4j.

- Data & RAG Pipelines: Design event-driven ingestion pipelines (Kafka/Spark) to chunk, embed, and index customer data into hybrid relational and vector databases.

- Multi-Tenant SaaS Execution: Implement strict tenant isolation across APIs, databases, and vector stores, alongside token-usage metering, RBAC, and subscription billing.

- Agentic Workflow Design: Architect and implement multi-step, stateful AI agents in Java capable of planning, tool/function execution, self-correction, and human-in-the-loop (HITL) approvals.

- Enterprise Tool Integration: Expose internal SaaS microservices, SQL queries, and third-party APIs safely as deterministic tools for LLM agents to invoke.

Under Qualifications:

Qualifications:

- 8+ years of production experience with Java (17/21) and the Spring ecosystem (Spring Boot, Spring Security).

- Hands-on experience building RAG pipelines, managing vector databases (pgvector, Qdrant, or Pinecone), and working with LLM APIs.

- Strong background in data engineering fundamentals (SQL optimization, ETL/ELT, Kafka, or Spark).

- Proficiency in modern frontend development using TypeScript and React/Next.js.

- Experience designing multi-tenant SaaS systems (OAuth2/OIDC, Stripe metering, Docker/Kubernetes).

- Hands-on experience building agentic architectures (e.g., ReAct, Plan-and-Execute, or multi-agent routing) using Spring AI, LangChain4j, or Temporal for durable workflow execution.

- Strong understanding of agent guardrails, state persistence, loop prevention, and sandboxing tool execution per tenant.

Interested candidates share their updated resume at shahmin@innovatixinc.com