Senior AI Engineer, Agents

Stack

Responsibilities

• ▹Invent the agent roadmap: work with sales, data and product to find repetitive expert work and produce a ranked list of agents worth building, with a view on feasibility, cost and impact
• ▹Build the agents yourself: architecture, tools, loops, sub-agents, memory, state and evaluation
• ▹Own the platform under the agents: a shared tool layer over EveryWatch data (pricing, auctions, listings, references, portfolios), an MCP surface over the existing backend, shared memory, tracing and a reusable eval harness
• ▹Harden WatchChat: multi-turn state and memory, latency, cost, tool-call reliability, regression gates
• ▹Make quality measurable: golden multi-turn datasets, programmatic verifiers for tool and argument correctness, LLM-as-judge on held-out sets
• ▹Treat cost and latency as design constraints: cheap models for routing and intent, strong models where they earn their keep, context budgets, caching and, where it pays off, fine-tuning (SFT/LoRA)

Requirements

• ▹Strong Python and LLM engineering skills
• ▹Creativity: you generate agent ideas the business had not thought of and can tell one that will work from one that only demos well
• ▹Strong architect: you can design an agent platform that is still standing in five years (state, memory, tool boundaries, sub-agent decomposition, evaluation, failure modes)
• ▹Proactive, business-minded approach

Soft skills

About the company

EveryWatch is the largest and most trusted data source in the secondary watch market, created by watch lovers. It serves collectors, vendors and enthusiasts under one roof.