We're a lean AI consultancy building the applied AI layer for a media and entertainment client. The client has proprietary foundation models. We're building data infrastructure, RAG systems, multi-agent orchestration, runtime, monitoring, and prototype AI agents on top of their model stack. We need a Technical Lead who can own the technical direction of this engagement across two phases. Phase 1: Discovery and Architecture Sprint (Weeks 1-4) This is where you earn the role. You work directly with the founder to: - Audit the client's model endpoints for production readiness (are they real APIs or research demos?) - Assess data readiness across media archives and content catalogs - Design the full system architecture (data pipelines, RAG, agent orchestration, runtime, deployment) - Map the dependency chain and produce a build plan with weekly milestones - Define success metrics, staffing needs, and handoff plan Identify risks and gaps before the build starts - You need to be the person who walks into an ambiguous technical landscape, figures out what's real, and comes back with a clear plan. Phase 2: Build Execution (Months 2-3+) With strong execution in the sprint, you move into leading the build: - Run daily standups and track progress across a team of 7-9 in US and India - Own technical decisions (vector DB, orchestration framework, approval flows) - Review PRs and architecture decisions for quality and consistency - Manage dependencies across data pipelines, retrieval, model gateway, orchestration, runtime, and monitoring - Coordinate with the client's India-based engineering team Keep async workflows tight with written decisions and clear handoffs - Flag scope creep before it becomes a problem The build phase is where the engagement extends. We expect 4-6+ months of build work following the sprint, and the Technical Lead is central to all of it. What this role is NOT: - Not a hands-on coding role. You evaluate and direct, not commit code daily. - Not a PM who tracks tickets but can't assess technical quality. - Not an architect who draws diagrams and disappears. You stay through execution. Must-haves: - 8+ years in software engineering, 3+ years leading technical teams or engagements - Hands-on production experience with RAG systems over real, messy data - Experience with multi-agent or LLM-powered application architecture (LangGraph, CrewAI, or custom orchestration) - Strong Python (you need to review it, not just read about it) - Experience leading distributed teams across timezones (especially US/India) - Strong written communication (decisions and handoffs must be documented clearly) - Comfortable walking into ambiguity, assessing what's real, and building a plan Strong pluses: - Media, entertainment, content, or IP-adjacent industry experience - Vector databases (Pinecone, Weaviate, Qdrant, pgvector) - Cloud deployment (AWS/GCP/Azure) with infrastructure-as-code - Data governance and audit requirements for enterprise AI - LLM evaluation/testing frameworks - Prior consulting or agency background To apply, send: - The most relevant engagement you've led (stack, team size, timeline, outcome) - Your specific experience with RAG and/or multi-agent systems in production - Your availability and rate