AI Engineer, Sr

• AI & Agent System Engineering Orchestration: Implement intent classification flows, tool routing logic, multi-step reasoning chains, session memory management, and structured output parsing. Tool Gateway: Build and maintain the deterministic validation layer checking LLM actions against policy rules. RAG Pipelines: Optimize document ingestion, chunking/embedding workflows, vector retrieval, and context injection. Guardrails & Testing: Implement "Trust Dial" enforcement in code to determine autonomous vs. user-confirmed actions. Build automated agent evaluation tests to catch regressions and hallucination risks. 2. Cloud-Native Service Development Scalable Architecture: Build containerized, stateless, horizontally scalable workloads designed to degrade gracefully if upstream banking APIs fail. Observability: Implement distributed tracing across agent calls and API hops, structured logging for audits, and metric instrumentation for latency/error rates. LLM Integration: Manage token budgets, streaming response handling, model fallback logic, and API error handling. Event-Driven Systems: Build idempotent consumers/producers on the Kafka pipeline with sub-second latency targets.

Desired Candidate Profile

Experience & Qualifications Essential Criteria Software Engineering: 4+ years of professional experience shipping and operating production systems at scale. Languages: Strong proficiency in Python and/or TypeScript with an emphasis on clean, testable code. Distributed Systems: Solid grasp of concurrency, fault tolerance, and API design (REST/GraphQL) in regulated environments. Agentic AI

Experience

Hands-on experience building/shipping LLM-based systems, RAG components, tool calling mechanisms, and AI safety features (hallucination detection, prompt injection defenses). Testing Rigor: Experience building automated evaluation pipelines to validate non-deterministic AI behavior. Cloud Native: Strong experience with containerized workloads, CI/CD pipelines, and cloud environments (AWS, Azure, or GCP).