We are seeking a technically deep and people-first AI Engineering Manager to lead the design, delivery, and responsible governance of enterprise-grade AI systems. You will own cross-functional teams building LLM applications, agentic workflows, multimodal pipelines, and ML platforms - while driving a culture of eval-first engineering, cost discipline, and production reliability at scale.

Key Responsibilities :

1. Technical Architecture & GenAI :

- Architect LLM applications across OpenAI, Claude, Gemini, and open-source models

- Design and govern MCP server/client architectures and agentic tool registries

- Build hybrid inference pipelines routing tasks to reasoning vs. fast models

- Manage thinking token budgets, verifier-generator patterns, and chain-of-thought evaluation

- Architect multimodal pipelines - vision-language, speech, document AI

- Optimize RAG pipelines - vector DBs, embedding strategies, retrieval tuning

- Lead prompt engineering, RAG evaluation (BERTScore, LLM-as-judge), and hallucination tracking

- Implement guardrails, safety filters, and compliance frameworks

2. Agentic Systems & Memory :

- Lead multi-agent orchestration using LangGraph, AutoGen, and MCP-native patterns

- Design long-term agent memory - episodic, semantic, and procedural layers

- Define human-in-the-loop controls and safety boundaries for autonomous agents

- Establish agent interoperability standards across internal and third-party systems

- Own security posture for agentic systems - prompt injection, tool misuse, scope creep

3. MLOps & Platform Engineering :

- Establish CI/CD for ML models - Argo, GitHub Actions, Tekton

- Scalable inference with Kubernetes, vLLM, Ray, Triton

- Model optimization - quantization, LoRA fine-tuning, distillation for production

- Evaluate and deploy SLMs for edge / on-device use cases

- Experiment tracking - MLflow, Weights & Biases; feature stores with Feast

- Observability - OpenTelemetry, Prometheus, Grafana; drift and bias monitoring

- Oversee synthetic data generation pipelines and data flywheel strategy

4. Eval-First Engineering Culture :

- Champion eval-driven development as a hard deployment gate

- Build team capability to write domain-specific evaluation suites

- Track regression benchmarks across model versions and prompt changes

- RAG evaluation using BERTScore, LLM-as-judge, RAGAS frameworks

- Implement AIOps - prompt caching, semantic caching, token budget governance

- Define cost-per-inference targets and track AI infrastructure spend

5. Team Leadership & Delivery :

- Lead and mentor AI/ML, Data Engineering, and MLOps engineers

- Own end-to-end delivery from requirements through production monitoring

- Drive Agile, DevOps, and MLOps practices across the team

- Govern AI coding assistant adoption and measure developer productivity lift

- Conduct code and design reviews; enforce architectural standards

- Build a high-performance, psychologically safe engineering culture

6. Governance & Responsible AI :

- Maintain EU AI Act documentation - Articles 11 & 13 conformity assessments

- Conduct high-risk AI assessments and manage incident reporting obligations

- Align with NIST AI RMF 2.0 and ISO/IEC 42001

- Implement responsible AI - fairness, explainability, privacy controls

- Translate business requirements into AI solutions for non-technical stakeholders

- Partner with Product, Architecture, Legal, and Security

Experience Required :

- 10+ Years in software / data / AI 5+ Years in AI/ML or GenAI systems 3+ Years leading engineering teams 1+ Years with agentic / LLM-native production

Required qualifications :

- Bachelor's / Master's in Computer Science, AI, Data Science, or equivalent

- Hands-on with OpenAI, Anthropic, Azure OpenAI, or open-source LLMs in production

- Proficiency in Python; familiarity with Go or TypeScript a plus

- Deep understanding of RAG, vector databases, and embedding strategies

- MCP architecture and agentic tool-calling standards

- Reasoning model operations and hybrid model routing (fast vs. thinking models)

- Multimodal pipeline experience - vision, audio, document AI

- MLOps tooling - MLflow, W&B, Argo, Feast

- EU AI Act literacy and NIST AI RMF 2.0 / ISO 42001 familiarity

- Kubernetes and distributed inference (vLLM, Ray, Triton)

- Eval frameworks - Braintrust, LangSmith, Inspect, or equivalent

- AIOps - prompt caching, semantic caching, token budgeting

Preferred / bonus :

- Managed AI coding assistant governance across an engineering org

- Published evals, benchmarks, or open-source AI tooling

- Experience leading AI red-teaming or safety functions

- AI incident response and post-mortems (hallucination events, agent misuse)

- Synthetic data generation and data flywheel design experience

- ISO/IEC 42001 implementation or audit experience

Technical Skills :

- Python / TypeScript MCP architecture & agentic tool registries

- LangGraph / AutoGen Reasoning models (o3, Claude, DeepSeek R1)

- OpenAI / Azure OpenAI / Anthropic Claude Multimodal AI (vision, audio, document)

- RAG pipelines / Vector DBs

- Kubernetes / vLLM / Ray / Triton Agent memory architecture (episodic, semantic)

- LoRA / QLoRA / quantization

- MLflow / Weights & Biases Braintrust / LangSmith / Inspect (evals)

- Argo / GitHub Actions / Tekton AIOps / semantic & prompt caching

- OpenTelemetry / Prometheus / Grafana Synthetic data pipelines / data flywheel

- Feature store (Feast) EU AI Act / NIST AI RMF 2.0 / ISO 42001