AI Engineer – Multi‑Agent Systems for Media Trading

About the role

Delve Deeper is a performance media agency serving the charity and nonprofit sector. The AI Engineer will design, build, and maintain a multi‑agent AI system that automates performance media trading decisions, ensuring high‑quality data handling and production reliability.

Key responsibilities

• Build, iterate, and maintain AI agents, handling prompt design, tool wiring, context routing, failure handling, and output validation.
• Integrate data from media platforms, normalize to schema, and route to the appropriate agent context while monitoring schema drift.
• Develop under an evaluation‑first approach: define success criteria, run benchmark tests, and track output quality across model versions.
• Design, implement, and sustain ETL/ELT pipelines for daily callouts and weekly optimisation reporting.
• Extend and operate the MCP connector library for external platform APIs, managing rate limits, retries, and resilient failure handling.
• Create Slack‑based human‑in‑the‑loop workflows for approvals, feedback capture, and exception alerts.
• Own production reliability of agents, monitor output quality, respond to incidents, and drive root‑cause fixes.

Required profile

• 4+ years of experience in software, data engineering, ML, or AI platform work with production ownership.
• Hands‑on experience building or operating LLM applications, agentic systems, or tool‑calling workflows.
• Strong Python and SQL skills for production‑grade code.
• Experience with media platform APIs such as Google Ads, Meta, DV360, Semrush, or SerpAPI.
• Proficiency with workflow orchestration tools (Airflow, Dagster, Prefect, dbt) and cloud environments (AWS, GCP, Azure) using containerised deployments.

Required skills

• Python
• SQL
• AWS
• GCP
• Azure
• Docker
• Airflow
• Dagster
• Prefect
• dbt
• MCP connector integration
• LLM application development
• Agentic system design
• Slack API and webhook automation
• Vector databases and RAG patterns
• ETL/ELT pipeline design
• Media platform APIs (Google Ads, Meta, DV360, Semrush, SerpAPI)

What we offer

• Hybrid work model: three days onsite in Minsk, two days remote.
• Competitive salary with annual bonus opportunities.
• Health and dental insurance after trial period.
• 28 calendar PTO days plus flexible sick‑day policy.
• Brand‑new office in Minsk and support for English tuition.
• Generous employee referral bonuses and mental‑health reimbursement.