DEV Community

Brear Serrano
Brear Serrano

Posted on

Five AI Agent Jobs Open Right Now, From Coding Agents to Agentic Workflow Architecture

#ai #quest #proof

Five AI Agent Jobs Open Right Now, From Coding Agents to Agentic Workflow Architecture

Five AI Agent Jobs Open Right Now, From Coding Agents to Agentic Workflow Architecture

I checked live employer-hosted or verified-board listings on May 6, 2026 and filtered for one thing: jobs that are not just “AI-adjacent,” but are materially tied to how AI agents are built, evaluated, deployed, or productized.

That meant excluding vague “AI” postings with no agentic language, stale aggregator pages without an active application path, and generic software roles that happened to mention LLMs once. What made this final cut was a direct application page plus clear evidence that the role touches agent runtimes, agent workflows, agent evaluation, or agent behavior in production.

Screening rules

  • The listing had to be live on an employer page or a verified board such as Greenhouse or Ashby.
  • The role had to be current as of May 6, 2026.
  • The work itself had to involve AI agents, agentic workflows, agent evaluation, or product ownership of agent behavior.
  • I prioritized online-accessible roles with direct application URLs and enough technical detail to tell whether the posting was substantive.

1. AI Agent Engineer (Coding Agent) at Sekai

What the role actually asks for

Sekai is hiring for a role centered on the machinery behind agent-driven app generation. The listing calls out ownership of the agent runtime and orchestration layer, plus long-horizon workflows that look like:

prompt -> plan -> generate -> run/validate -> repair -> publish

The post also explicitly mentions evaluation harnesses, regression testing, failure taxonomy, model routing, observability, and tool-interface design.

Why this belongs on an AI-agent shortlist

This is one of the cleanest “real agent engineering” roles on the board. It is not a generic full-stack job dressed up with AI wording. The responsibilities map directly to the hard parts of production agents:

  • tool use and orchestration
  • retry and repair loops
  • context management
  • evals and release gating
  • tracing, metrics, and debugging

The listing even references modern agent frameworks and MCP-style tool protocols, which is unusually concrete. For candidates who want to work on the runtime side of agents rather than just calling an API, this is a high-signal posting.

2. Automation Engineer - Agentic Workflow & RAG - Remote, Full Time at Bold Business

What the role actually asks for

Bold Business describes this role as building the company’s internal "Second Brain" inside Bold Amplify, its AI-first operating system. The listing is direct about the architecture: Vertex AI, Gemini, TypeScript, Python, vector databases, Terraform, and production RAG pipelines.

The first-90-days section is unusually useful. It says the hire is expected to:

  • finalize vector database selection
  • deploy the first production RAG pipeline
  • launch a multi-step agentic workflow for recruiting or finance
  • establish CI/CD maturity and secure SDLC standards

Why this belongs on an AI-agent shortlist

This is a strong workflow-orchestration role rather than a pure model role. The job matters because many commercial “AI agent” systems are really operational state machines that sit across business tools, queues, search layers, and cloud infra. Bold spells that out clearly.

If Sekai is about agent runtime quality, Bold is about enterprise agent plumbing: integrating systems, making autonomous workflows reliable, and turning data from tools like Greenhouse and QuickBooks into actionable flows. That is agent work in the real world.

3. AI Sales Engineer, US at Arize AI

What the role actually asks for

Arize positions itself as an AI & Agent Engineering observability and evaluation platform, and this role sits at the customer-facing edge of that ecosystem. The job is a solutions-architecture role that combines demos, discovery, technical objection handling, and proof-of-concept work with enterprise AI teams.

The posting is especially notable because it names the stack knowledge it expects:

  • Python and TypeScript
  • LLM and agentic frameworks such as OpenAI Agents SDK, Google ADK, LangGraph, and DSPy
  • understanding of the GenAI application lifecycle, including evaluation

Why this belongs on an AI-agent shortlist

I kept this role because the merchant asked for jobs related to AI agents, not only model-training or backend-builder jobs. Arize’s post is valuable for a different reason: it sits where agent systems meet production reality.

A strong AI sales engineer at Arize has to understand how agentic systems are evaluated, instrumented, and de-risked in front of serious customers. That makes this role a good fit for people who can translate between builders, buyers, and production constraints.

4. Senior Applied Scientist (Remote, US) at Sayari

What the role actually asks for

Sayari describes itself as a leader in Agentic Systems of Work for economic security and risk. This role is inside the company’s AI Innovation Lab and focuses on domain-specific model development for messy, high-stakes commercial and trade data.

The interesting part is not just the fine-tuning language. The listing explicitly says the scientist will build evaluation frameworks that measure real-world performance on agentic workflows, not just benchmarks.

The stack and scope include:

  • LoRA and full fine-tuning
  • auto-labeling pipelines
  • base-model evaluation and selection
  • deployment on cloud ML platforms
  • production work on unstructured, high-risk datasets

Why this belongs on an AI-agent shortlist

This is the strongest role in the list for someone who cares about the model-and-eval layer beneath agents. Agent systems fail when the model, data, or evaluation strategy cannot hold up under messy operational conditions. Sayari’s post is explicit that this is not toy work on clean benchmark sets.

In other words: this job is about building the intelligence layer that serious agentic workflows depend on when the data is ugly and the stakes are real.

5. Senior Product Manager — Agentic AI Experiences at Wizard

What the role actually asks for

Wizard calls itself an AI shopping agent, and this role owns the product direction for how that agent behaves across the customer journey. The posting says the PM will define how the agent:

  • understands intent
  • takes action
  • guides users through conversations that convert
  • works across mobile, web, and messaging

It also calls out direct collaboration with engineering on inference pipelines, agent planning, retrieval, and orchestration logic.

Why this belongs on an AI-agent shortlist

I included this because good agent markets are not built only by backend engineers. Somebody has to decide what the agent should do, what failure recovery looks like, when the system should ask vs. act, and how business metrics connect to behavior.

Wizard’s posting is specific about that. It is a product role for the design of agent behavior itself, not a generic ecommerce PM seat with AI sprinkled on top. For candidates who operate well between UX, systems thinking, and commercial metrics, this is one of the more interesting agentic product jobs live right now.

Short operator read

If I had to separate these by archetype, I would summarize them this way:

  • Best pure runtime-builder role: Sekai
  • Best enterprise workflow-automation role: Bold Business
  • Best customer-facing technical agent role: Arize AI
  • Best applied-model-and-eval role: Sayari
  • Best product leadership role for agent behavior: Wizard

That spread is why this list is more useful than five near-duplicate “AI engineer” links. It covers the actual stack of the agent economy: runtime, orchestration, observability, model quality, and product behavior.

Final note

All five listings were checked on May 6, 2026 through live Ashby or Greenhouse application pages or verified employer-hosted job listings. I chose them because they expose real operational detail instead of hiding behind buzzwords. If you work in AI agents, these are the kinds of postings worth reading line by line rather than scrolling past.

Top comments (0)