Senior AI Engineer

TMTG•Sarasota, FL

2d•Remote

About The Position

You'll work directly with the platform architect and other developers to bring deep, hands-on LLM-stack expertise: you've built these systems before, you know where they break, and you know what "good" looks like in production. The work is greenfield. You won't be maintaining someone else's pipeline — you'll be standing up the core AI infrastructure for a platform with millions of active users, and shaping the engineering practices around it as the team grows.

Requirements

5+ years of backend engineering experience in a statically typed language (Go, Java, Kotlin, C#)
Shipped production LLM-backed features — retrieval-augmented generation, streaming responses, tool use — and lived with them after launch
Hands-on experience with the LLM serving stack: routing across multiple model providers, failover, token streaming, and cost/usage metering
Experience building retrieval systems: vector search, embedding pipelines, context assembly, and citation-backed answers
Experience with failure modes: hallucination, retrieval misses, provider outages, cost blowouts — and building the instrumentation to catch them
Pragmatic about evaluation — knowing how to measure whether AI answers are actually good (relevance, safety, source quality) with simple, repeatable tests, not just academic benchmarks
Strong API design instincts; comfortable owning a service end to end, from schema to deploy to on-call
US-based and authorized to work in the United States

Nice To Haves

Go (strongly preferred)
Python
Experience with LLM gateways or serving infrastructure (LiteLLM, vLLM, TGI, or similar)
Vector databases (Qdrant, pgvector) and embedding pipelines
Fine-tuning open-weight models (LoRA or full fine-tunes) and the eval discipline that goes with it
Content moderation or safety tooling experience
Familiarity with Ruby on Rails or React/TypeScript (you'll integrate with both)

Responsibilities

Building AI-based features into a platform serving millions of users.
Bringing deep, hands-on LLM-stack expertise.
Building production LLM-backed features — retrieval-augmented generation, streaming responses, tool use — and lived with them after launch.
Hands-on experience with the LLM serving stack: routing across multiple model providers, failover, token streaming, and cost/usage metering.
Building retrieval systems: vector search, embedding pipelines, context assembly, and citation-backed answers.
Thinking in failure modes: hallucination, retrieval misses, provider outages, cost blowouts — and building the instrumentation to catch them.
Pragmatic about evaluation — knowing how to measure whether AI answers are actually good (relevance, safety, source quality) with simple, repeatable tests, not just academic benchmarks.
Designing APIs and owning a service end to end, from schema to deploy to on-call.