Applied AI Engineer

Job Summary

A seed-stage startup building a plug-and-play contact center platform that delivers personalized customer support through trained on-demand agents.

Design experiments to determine whether changes to AI systems actually improve response quality. Build evaluation harnesses, datasets and failure taxonomies that reflect real-world distributions rather than cherry-picked queries. Diagnose AI system regressions and identify underlying classes of failures rather than isolated instances. Investigate retrieval failures and determine how to fix the broader class of retrieval problems. Reason about where system behaviour should be deterministic versus LLM-driven. Explore how AI systems should remember information over long-running customer relationships. Argue for architectural decisions using evidence, with experiments and real production feedback informing what gets shipped.

Role Overview

We are looking for an AI Engineer / Applied AI Engineer with 3-6 years of experience to work on AI-powered customer support systems for D2C e-commerce brands. The role is focused on building and improving AI systems that handle real customer conversations every day. This role leans more toward applied research than conventional backend engineering. You ll work on evaluation, retrieval, agent architectures, reasoning, memory and AI system failure analysis while designing experiments and making architectural decisions based on evidence.

Responsibilities

• Design experiments to determine whether changes to AI systems actually improve response quality.
• Build evaluation harnesses, datasets and failure taxonomies that reflect real-world distributions rather than cherry-picked queries.
• Diagnose AI system regressions and identify underlying classes of failures rather than isolated instances.
• Investigate retrieval failures and determine how to fix the broader class of retrieval problems.
• Reason about where system behaviour should be deterministic versus LLM-driven.
• Explore how AI systems should remember information over long-running customer relationships.
• Argue for architectural decisions using evidence, with experiments and real production feedback informing what gets shipped.