About KPI Partners:

KPI Partners is a data, analytics, and AI consultancy founded in 2006. We have 600+ consultants and 300+ enterprise customers, headquartered in Oakland, California with delivery and go-to-market teams across India, Mexico, and the UK. We build modern data platforms, analytics products, and production-grade AI systems on AWS, Azure, GCP, Databricks, and Snowflake.

The Role:

You will be the technical face of our Agentic AI practice with customers. You will shape and sell agentic solutions to enterprise clients: finding the business processes where autonomous agents can take over multi-step work, and turning them into sound architectures, roadmaps and delivery plans. This is a deeply technical, customer-facing leadership role.

Key Responsibilities:

- Lead discovery with client executives and process owners to find workflows ripe for autonomy.

- Design end-to-end agentic architectures: single- and multi-agent orchestration, planning and reasoning, tool and API access, memory and state.

- Own the technical narrative in proposals, SOWs and orals: solution approach, autonomy levels, estimates, staffing, risks and assumptions.

- Set reference architectures, reusable accelerators and best practices for agentic systems: agent design patterns, orchestration, tool governance, evaluation and AgentOps / LLMOps.

- Hire, mentor and lead a team of agentic AI architects, AI engineers and solution leads.

Required Qualifications:

- 15+ years in technology, with 8+ years in data, analytics, ML or AI and 2 - 3+ years leading GenAI / LLM initiatives.

- Proven customer-facing solutioning and experience in a consulting, systems integrator, hyperscaler or ISV field role.

- Deep conceptual mastery of agent fundamentals: reasoning and planning patterns, tool and function calling, memory and state management.

- Executive presence: comfortable presenting to C-level audiences and turning technical depth into business value.

- Bachelor's degree in Computer Science, Engineering or a related field; Master's preferred.