AI Agent Engineer

About the Role

At wearemighty, we’re building at the intersection of AI, product, and real-world utility.

Our core product, trysecretsauce.ai, is an AI-powered platform that helps businesses generate high-quality marketing content while staying fully aligned with their brand identity.

We’re hiring AI Agent Engineers, engineers who work one layer above the model.

This is not a traditional backend or ML role. You will not be training foundation models or tuning gradients. You will be building the systems around the model that make it actually useful in production, including agent loops, tool integrations, context pipelines, retrieval systems, evaluation harnesses, and the cost and latency layers that keep everything fast and scalable.

The term “AI engineer” has become overloaded, so to be specific, we are not hiring an ML engineer. We are hiring someone whose craft is turning “the model works in a notebook” into “the product works reliably for thousands of users every day.”

You will work closely with product and engineering teams to design and ship agentic systems that power real customer workflows end-to-end, with a strong focus on reliability, cost efficiency, and user experience.

We are an equal opportunity employer. We welcome applications from candidates of all backgrounds and are committed to building an inclusive environment.

Your Responsibilities Will Include

Designing and shipping production-ready AI agent systems that power workflows inside trysecretsauce.ai
Building and refining agent loops, including planning, tool use, reflection, and retry strategies
Owning prompt engineering and context engineering as core disciplines, including structured prompts, versioning, template libraries, and few-shot strategies
Architecting retrieval systems using RAG approaches, including hybrid search, reranking, chunking, and document pipelines
Designing memory systems, including short-term task memory and long-term user or agent memory across sessions
Building and maintaining evaluation harnesses, including benchmarks, A B testing, golden datasets, and regression suites
Optimising cost and latency using techniques such as prompt caching, KV reuse, streaming, model routing, and response caching
Building agent infrastructure, including sandboxes, tool wrappers, MCP servers, permission layers, and observability
Taking features from prototype to production, including logging, error handling, rate limits, and fallback strategies
Collaborating closely with product and stakeholders to ship reliable, user-facing AI features
Contributing to system design decisions and engineering best practices

As a Successful Candidate, You Would Have

Production experience building LLM-powered or agent-based applications

Hands-on expertise in at least three of the following

Agent architectures
Prompting or context engineering
Tool use
Retrieval or RAG systems
Evaluation design
Memory systems
Prompt or response caching
Strong software engineering fundamentals, including API design, concurrency, testing, and debugging production systems
Fluency in TypeScript is required, and working knowledge of Python is preferred
A strong bias toward empirical, metrics-driven development, including evaluation, latency, and token cost optimisation
Comfort operating in the space between model behaviour and product UX

Strong Candidates May Also Have

• Experience designing multi-agent or long-horizon agent systems
• Experience building model routing, prompt caching, or KV caching infrastructure
• Experience developing evaluation platforms or eval-as-code tooling
• Contributions to open-source agent frameworks such as LangGraph, CrewAI, AutoGen, or the MCP ecosystem
• Experience with fine-tuning, RLHF, or distillation, which is useful but not required

What This Role Is Not

Not a research role. We are not training foundation models
Not a traditional ML engineering role. You will not be building training pipelines or classical ML systems
Not a pure backend role. This role sits between model behaviour and product experience

What We Look For

You have shipped real systems in production and understand reliability at scale

You care about performance, cost, and user experience equally

You take ownership and think in systems, not just features

You are pragmatic and comfortable working in fast-moving environments

If you are interested in building production-grade AI systems that power real-world marketing workflows through trysecretsauce.ai, we would like to hear from you.

Even if your experience does not match every requirement, we encourage you to apply.