About the role
Role & responsibilities
AI Architecture & Platform Leadership
- Define end-to-end AI architecture: data training evaluation deployment monitoring governance.
- Establish reference architectures for:
- RAG at scale (hybrid retrieval, reranking, caching, evaluation harnesses, feedback loops)
- Agentic AI systems (ReAct / plan-and-execute patterns, tool calling, memory, multi-agent collaboration)
- LLM serving (latency/cost optimization, model routing, caching, safety layers)
- Drive standardization: prompt/version management, evaluation pipelines, model release gates, and incident response.
Database + Enterprise Data Architecture for AI
- Define patterns for integrating databases into AI solutions (secure access, performance-aware querying, data minimization, auditability).
- Guide teams on data modeling considerations for AI use cases (feature extraction, retrieval metadata, lineage).
Python Middleware / Integration Architecture (Intermediate Server Layer)
- Define and govern architecture for Python middleware services (FastAPI) that orchestrate:
- deterministic business logic
- ML/LLM inference
- external LLM provider calls (e.g., OpenAI)
- DB interactions and enterprise integrations
- Set engineering standards for reliability: retries/timeouts, circuit breakers, caching, observability, secure secrets management, and API governance.
Deterministic AI + Cost Governance
- Establish decision frameworks for when to use deterministic logic vs smaller domain LLMs vs large external LLMs, balancing:
- cost, latency, accuracy, explainability, and regulatory constraints
- Promote LLM where needed” design, including fallback strategies (deterministic-first, LLM escalation, human-in-the-loop where required).
MLOps / LLMOps at Scale
- Build/own MLOps maturity: experiment tracking, model registry, lineage, CI/CD, monitoring, drift detection.
- Define operational SLOs/SLAs for AI services (availability, latency, cost, quality).
Responsible AI, Security & Governance
- Implement Responsible AI practices: fairness, interpretability, auditability, privacy-by-design.
- Lead GenAI security: prompt injection defenses, data leakage prevention, access control, red teaming, compliance.
Technical Leadership & Stakeholder Management
- Mentor AI/ML engineers and data scientists; run design reviews and architecture governance.
- Translate business goals into AI roadmaps, KPIs, and delivery plans.
- Evaluate vendors/tools and decide build vs buy (open-source vs managed LLM platforms).
Required Qualifications
- 10–15+ years overall experience with 7+ years in ML/AI systems delivery.
- Proven record of delivering production AI systems (not only POCs), including GenAI.
- Strong expertise in:
- LLMs, RAG, evaluation, and agentic AI patterns
- AI system architecture and cloud-native design
- MLOps/LLMOps, monitoring, reliability engineering
- Strong software engineering background (APIs, distributed systems concepts, security fundamentals)
- Experience designing systems that integrate databases with AI services (directly or via data/platform layers).
- Strong hands-on understanding of Python service architecture (FastAPI or similar) for orchestration/integration.
Preferred / Nice-to-Have
- Experience with Banking / Wealth Management domain LLMs (smaller/specialized models), and governance for regulated environments.
- Distributed ML / performance optimization (GPU inference, quantization, vLLM/Ollama).
- Knowledge graphs + retrieval (Neo4j), multimodal systems, speech/CV pipelines.
- Exposure to advanced areas: reasoning/planning, secure AI, edge AI, robotics/RL, quantum ML (as applicable).
- Architecting AI Systems end-to-end architecture patterns, scalability, resiliency, governance
- Agentic AI Systems / Advanced Reasoning and Planning autonomous agents, planning, evaluation, safety
- MLOps / Software Engineering for ML / API-driven Cloud Native Solutions production-grade AI platforms
- Distributed Machine Learning / Parallel & Distributed Programming scale-out training/inference patterns
- Fair, Interpretable, Trustworthy ML + Security topics Responsible AI + GenAI security posture
- Multimodal AI / Multimodal Information Retrieval cross-modal enterprise AI solutions
- AI & ML Techniques for Cyber Security adversarial ML awareness, secure AI operations