Senior AI Engineer – Agentic Systems & Cloud Platform

About the role

Nysonian is looking for a senior AI engineer to own the end‑to‑end design, production, and operation of agentic AI systems that power its commerce platform and internal CRM. This is a builder role with full architectural ownership, not just prompt tweaking.

Key responsibilities

• Design, ship, and maintain multi‑agent orchestration layers, tool‑calling schemas, and human‑in‑the‑loop escalation for live customer workflows.
• Build and operate the tool layer (MCP servers) and integrations with Shopify, ShipHero, Gorgias, Klaviyo, telephony, and internal APIs.
• Run a multi‑model stack, routing jobs by cost and latency, and evaluate new model releases in production.
• Implement guardrails for hallucination, PII handling, prompt‑injection defense, and financial action safety.
• Create evaluation and observability pipelines, including golden datasets, regression suites, LLM‑as‑judge scoring, and dashboards.
• Develop the RAG layer with chunking, embeddings, hybrid retrieval, and vector‑store hygiene.
• Extend production services on AWS and the MERN stack, ensuring CI/CD, versioning, staged rollouts, cost tracking, and on‑call ownership.

Required profile

• 4+ years of cloud and production software engineering experience with full post‑deployment ownership.
• Hands‑on experience shipping agentic AI systems that interact with real users, including tool‑calling and structured‑output validation.
• Deep knowledge of multi‑agent orchestration, state management, async/event‑driven design, and cost/latency engineering.
• Strong evaluation discipline and ability to size experiments and measure impact reliably.

Required skills

• AWS services (EC2, Lambda, S3, RDS, IAM, VPC, queues)
• Containers, CI/CD, infrastructure‑as‑code, logging/metrics (Prometheus/Grafana)
• MERN stack (MongoDB, Express, React, Node.js) and REST/GraphQL integrations
• Tool integration experience with Shopify, ShipHero, Gorgias, Klaviyo, telephony APIs
• LLM engineering at scale: prompt engineering, guardrails, security (OWASP LLM Top 10), fine‑tuning (LoRA/QLoRA)
• Evaluation frameworks, golden datasets, regression testing, observability dashboards