Applied AI Engineer

This is a fully remote position, open to applicants in Portugal.

📋 Description

• Develop and manage a production Slack-native SQL BI analyst agent that converts natural-language business inquiries into governed SQL.

• Validate queries, ensure result accuracy, and provide referenced evidence for every statistic.

• Create executive P&L responses, daily health reports, and data-driven root cause analyses.

• Enhance agents for customer-facing intent triage, routing, RAG responses, conversational context, localized brand voice, and escalation processes.

• Integrate agents with helpdesk and CRM systems via webhooks, session lifecycle management, intent tagging, and automated escalation tickets.

• Develop risk-stratified tools for back-office APIs that include validation, confirmation workflows, and controlled autonomy.

• Safeguard agents against prompt injection, tool misuse, and data exfiltration.

• Construct agents utilizing LangGraph, Anthropic Agent SDK, MCP, or comparable frameworks.

• Engineer feedback loops, semantic memory, decision audit logging, evaluation harnesses, and regression suites.

• Create and implement models for churn, lifetime value, bonus sensitivity, player risk, collusion, bot play, multi-accounting, and treasury/payment anomalies.

• Deploy governed ML signals with versioning, SLAs, freshness, drift, calibration monitoring, and automated retraining.

• Manage multi-vector withdrawal risk scoring and evidence-aware re-scoring.

• Convert policies into deterministic, configurable, auditable rules; simulate and backtest modifications.

• Design holdouts and control groups, assess uplift, and conduct in-depth analyses.

• Develop agent supervisor and approval interfaces, review queues, session replay, grading modules, and model/evaluation datasets.

• Create dashboards for decision audits, agent performance, risk review queues, and KPIs.

• Report directly to the Head of Data and collaborate with AI, BI, Infrastructure, Customer Success, and product teams.

⛳️ Requirements

• A minimum of 4 years' experience in software, data science, or machine learning engineering.

• At least 1 year of experience building LLM-powered agents in a production environment.

• Familiarity with tool usage and function calling, structured outputs, retrieval and memory, and multi-step orchestration.

• Experience with LangGraph, Anthropic Agent SDK, MCP, or similar orchestration frameworks.

• Background in production RAG systems, including grounding, chunking, retrieval quality, hallucination control, and refusal strategies.

• Understanding of prompt injection, tool-call abuse, and data leakage defenses.

• Ownership of the production ML lifecycle, including feature engineering, training, serving, monitoring, and retraining.

• Experience in fraud, risk, or abuse detection is highly desirable.

• Statistical rigor in experimental design, including holdouts, control groups, uplift measurement, and score calibration.

• Experience in constructing evaluation harnesses and regression suites for non-deterministic systems.

• Proficient in Python.

• Comfortable working with TypeScript.

• Strong SQL skills, particularly with columnar analytical databases; ClickHouse is preferred.

• Capability to build stakeholder-ready interfaces using a front-end framework or tools like Streamlit.

• Experience designing systems where model outputs facilitate deterministic execution.

• Familiarity with LLM observability and tracing tools, such as Langfuse or LangSmith.

• Nice to have: experience in iGaming or high-trust transaction-intensive environments.

• Nice to have: knowledge of helpdesk or customer service platform integration.

• Nice to have: experience with blockchain or crypto-native transaction flows.

• Nice to have: expertise in constrained optimization, bandits, or reinforcement learning.

• Nice to have: knowledge of rule engines, decision management systems, and Slack app development.

• Nice to have: experience with Kafka or MSK consumers, idempotent processing, and failure handling.

🏝️ Benefits

• Fully remote working environment.

• Asynchronous-first communication culture.

• Preference for EU timezone overlap.

• High level of autonomy and ownership over domain decisions.

• Architecture decisions are documented and actively debated.

• Opportunity to work in a global, multi-tenant scale engineering environment.