Production AI Engineer: Build Real-World AI Agents (Equity)

Unify; New York (NY); Sur place; USD 250 000 - 320 000; Plein temps

Avantages offerts par ce poste

Equity

Résumé du poste

Unify is building the first AI-native outbound platform where agents and sellers work side by side. As an Applied AI Engineer, you will bring agent capabilities from zero to one and own the harnesses that turn model capabilities into production actions.

You will work closely with product, design, and engineering to build agents that research accounts, identify opportunities, and execute GTM workflows. You’ll move quickly from prototype to customer usage, then use evaluations and real execution

Qualifications

• 3+ years of software engineering experience shipping LLM applications or agents.
• Built systems with tool calling, multi-step execution, and evaluations.
• Strong programming in TypeScript or Python; backend systems and API integration.

Responsabilités

• Design the execution loops, tool interfaces, context, and memory for agent harnesses.
• Ship AI products end to end, connecting agents to data, integrations and workflow infrastructure.
• Build evals from real customer tasks and use traces to improve agent behavior.
• Own reliability, latency, and cost; ensure production readiness and observability.
• Experiment with models, prompting strategies, retrieval approaches, and orchestration patterns.
• Shape reusable tools and engineering practices with product and design.

Connaissances

LLM apps

Tool calling

Multi-step execution

TypeScript

Python

API integration

Outils

Backend systems

Evaluations

Description du poste

Unify is building the first AI-native outbound platform where agents and sellers work side by side. As an Applied AI Engineer, you will bring agent capabilities from zero to one and own the harnesses that turn model capabilities into production actions.

You will work closely with product, design, and engineering to build agents that research accounts, identify opportunities, and execute GTM workflows. You’ll move quickly from prototype to customer usage, then use evaluations and real execution