Senior AI/ML Engineer

📋 Job Overview

We're forming a dedicated delivery squad for a high-growth international technology client. The goal is to move customer support from reactive to proactive. The new platform will detect service issues from the client's operational data, assess their business impact, trigger remediation, and let affected customers know before they have to ask.

🏢 About Zinkworks

Zinkworks partners with leading Telecommunications and Financial Services organizations to modernize legacy systems, migrate mission-critical platforms to the cloud, and engineer AI-driven automation. From OSS transformation to rApp development and network intelligence, our teams simplify complexity and turn it into competitive advantage. Based in Ireland and operating across the EU, UK, and US, Zinkworks combines deep domain expertise with delivery excellence to help clients modernize faster and operate smarter.

🎯 The Role

You'll own the intelligence at the heart of the platform: the models that detect service outages and degradation and assess how serious they are for the business. You'll start with anomaly detection on operational telemetry and carry the work through to production MLOps and, later in the programme, an LLM-based agent with retrieval and tool use.

✅ Key Responsibilities

• Explore historical telemetry and operational data, assemble labels and engineer features, including the baselines that detection depends on.
• Build a statistical baseline detector alongside an ML anomaly model, then evaluate, threshold and tune both against shadow-mode output before they go live.
• Serve models in near real time and instrument them with model monitoring and shadow-mode tracking.
• Design and train models that detect partial service degradation at a granular level, judged against expected performance rather than fixed thresholds.
• Upgrade the initial outage detector to a supervised model using labels generated in production.
• Build MLOps foundations: a model registry, retraining triggers, champion/challenger comparison, drift detection and automated evaluation, which are the preconditions for letting models act without human review.
• Develop an internal AI agent with retrieval, tool use and guardrails, supported by an evaluation harness, red-teaming and an explicit refusal policy.
• Work closely with data engineering on feature pipelines and with backend engineering on how model output feeds impact scoring and remediation.

📌 Required Qualifications

• Several years of hands-on experience taking ML models from exploration to production.
• Strong Python and the core data science stack (pandas, NumPy, scikit-learn or similar).
• Practical experience with anomaly detection or time-series modelling, including thresholding and dealing with noisy or sparse labels.
• Solid understanding of model evaluation, monitoring and drift in a live environment.
• Experience with MLOps tooling and practices: model registries, experiment tracking, automated retraining and CI/CD for models.
• Experience on a major cloud platform, ideally Google Cloud (e.g. Vertex AI, BigQuery).
• The ability to explain model behaviour and trade-offs clearly

⭐ Desirable Experience

• Experience building LLM applications with retrieval-augmented generation, tool use and guardrails, plus evaluation and red-teaming of those systems.
• Background in telecoms, networking or large-scale operational data.
• Experience with shadow deployment, canary rollout or champion/challenger testing of models.
• Familiarity with geospatial or hierarchical data.

🎁 Benefits

Competitive

🛂 Visa & Eligibility

Based in Ireland and operating across the EU, UK, and US, Zinkworks combines deep domain expertise with delivery excellence to help clients modernize faster and operate smarter.