CI&T - [Job-29167] Principal Architect

• Drive day-to-day delivery execution (planning, dependencies, unblocking, risk management) and ensure commitments are met. • Ensure engineering standards are applied (definition of done, testing strategy, code quality, documentation, operational readiness). • Provide clear engineering updates to stakeholders and escalate promptly when needed. • Enable team execution without direct line management: support onboarding and ways of working, mentor/coach engineers and tech leads, share best practices. • Ensure knowledge transfer and handover: document decisions, runbooks, and key architectural choices; facilitate smooth transition at the end of engagement. • Partner with Product/Business to identify where agentic AI provides clear value (workflow automation, assisted decision-making, content acceleration). • Translate needs into technical objectives (latency, cost, quality, robustness, compliance) and success criteria (KPIs, A/B testing, guardrails). • Design agentic architectures: tool orchestration, planning, memory management, context management, retrieval (RAG), routing, multi-agent patterns, human-in-the-loop. • Implement production patterns: prompt/versioning, evaluation harnesses, regression tests, feature flags, canary releases, and monitoring for quality/cost/latency. • Engineer for resilience: fallbacks, timeouts, retries, circuit breakers, safe tool execution, sandboxing, secrets management. • Implement guardrails: tool policies, content filtering, PII redaction, policy-as-code, access controls, auditability, and traceability (traces, conversations, decisions). • Partner with IT/Security/Cloud teams to ensure privacy, security, and risk compliance at scale. • Define and enforce quality gates before production (red-teaming, adversarial testing, bias, hallucination risk management). • Build and operate the ML value chain: data contracts, data quality, lineage, drift monitoring, dataset management, training/inference pipelines. • Own production operations for models and agentic services: deployment, scaling, observability, incident response, SLOs, post-mortems. • Industrialize continuous evaluation: offline evaluations, golden sets, human feedback loops, scorecards. • Coach tech leads and engineers on delivery habits, quality, and operational excellence (without direct people management). • Strengthen ways of working: documentation, testing, product ownership (“you build it, you run it”), and incident readiness. • Facilitate collaboration across software engineers, data engineers, data scientists, ML engineers, SRE, product, and security.