AI Engineer — Confidential AI-Native Travel & Concierge Platform

Aionia Group

New York, Northern (NY, KY)

Hybrid

USD 350,000 - 500,000

Full time

14 days+

Application generator

Turn this role into an interview — a resume and cover letter built around what this employer wants.

Get past ATS filters

Job summary

Aionia is seeking an AI Engineer in New York, NY to own the agent layer end-to-end, from prompts to tool calling and memory. The role blends full-stack engineering with production-grade AI systems in a hybrid, full‑time setting.

You will build and scale an AI-native travel concierge platform backed by a major global travel partner, shaping failure modes, provider strategy, and production reliability.

Qualifications

4+ years of experience as a fullstack or backend-leaning engineer with experience building on top of LLM systems.

Responsibilities

Own the conversational intelligence layer end‑to‑end: agent loop, prompt architecture, tool calling, state, memory recall, streaming.
Ensure reliability of multi-turn flows with refinement loops and graceful degradation.
Develop the tool/UI rendering layer between agent outputs and the product surface.
Define provider strategy across OpenAI, Anthropic, and Gemini including caching and structured outputs.
Build evaluation infrastructure for agent behavior.

Skills

LLM systems

TypeScript

Production agentic systems

AI-native workflows

Fast learner

OpenAI API

Education

Top-tier university degree

Tools

Tool calling

Memory handling

OpenAI API

Anthropic API

Gemini API

Job description

AI Engineer
Own the agent layer, end to end.

Confidential AI-Native Travel & Concierge Platform · New York, NY

New York, NY Full-time · Hybrid $350,000 – $500,000 + Competitive Equity Backed by a Major Global Travel Booking Platform · ~15 employees

About the Company

This confidential AI-native travel and concierge platform knows the places, prices, perks, and upgrades that make every experience better — and, most of all, it knows the traveler. Tell it what you're looking to do and it plans the trip, books the travel, and secures the reservation, with direct access to the world's largest travel and hospitality brands so every option is real, bookable, and priced to the minute.

The company is backed directly by a major global travel booking platform, with a lean 15-person team building the product today across AI, consumer, ecommerce, and education use cases.

You'll turn LLM capabilities into a product that consistently produces correct, useful actions — not just responses.

The Opportunity

You'll own the conversational intelligence layer end-to-end: agent loop, prompt architecture, tool calling, state, memory recall, streaming, and failure modes. This is a hands-on build role for an engineer who wants full ownership of how an AI product actually behaves in production — not just how it responds.

What You'll Own

Full agent architecture: prompt stack, tool calling, multi-step reasoning, streaming
Conversational state and memory plumbing — productionizing the existing three-pass pipeline
Reliability of multi-turn flows: refinement loops, error recovery, graceful degradation
Tool/UI rendering layer between agent output and the product surface
Provider strategy across OpenAI, Anthropic, and Gemini: caching, structured outputs, fallback
Eval infrastructure for agent behavior

Requirements

4+ years of experience as a fullstack or backend-leaning engineer, with experience building on top of LLM systems
Background in full stack, backend, or product engineering — not primarily ML research or model training
Experience with the full agent architecture: prompts, tools, memory
Can articulate specific failure modes debugged in production
Demonstrates high raw intelligence and ability to learn quickly
Aggressively uses AI-native workflows in day-to-day coding
Has shipped a production agentic system built on a model provider API (OpenAI, Anthropic, or Gemini) that real users interact with
Worked at a high-growth startup or a top tech company
Degree from a top-tier university or equivalent strong signal
Proficient in TypeScript (primary stack; Python is also fine)
Has significant side projects building AI/agentic systems

Traits to Avoid

Deep ML/research engineers whose primary output has been model tuning or static recommendations, not dynamic agentic behavior
Engineers from large, slow-moving companies with no independent side projects or AI-native habits
Candidates from startups that never gained real traction, without other strong experience elsewhere to offset it

Interview Process

30-minute conversation with the hiring team to discuss background, the role, and mutual fit.
Technical Interview

Deep-dive on agent architecture experience, production failure modes you've debugged, and how you think about tool calling and memory.
Onsite / Final Round

Meet engineering leadership. Walk through a practical scenario on agent reliability and provider strategy.

Role Details