Managing Director, AI Platform Architect

Baker Tilly•San Francisco, CA

About The Position

The Managing Director, AI Platform Architect is the founding architect of Baker Tilly's AI Product Group (APG) — the senior-most individual contributor in the organization and the single design authority for the product, agentic, and data architecture of everything APG builds. The architect designs the agent platform the firm's AI products run on (agent harness, orchestration, tool and MCP integration, retrieval, memory and context management, and the evaluation infrastructure that decides what ships), co-designs the intelligence substrate (ontology, knowledge graph, judgment layer) with the substrate lead, chairs APG's technical design authority, and serves as the CAITO's senior technical counsel. This is a hands-on, build-first role: the architect prototypes the patterns product teams then harden, sets the technical bar for the founding engineers, and leads the deliberate absorption of architecture from the firm's expert engineering partners so Baker Tilly owns its platform rather than renting it.

Requirements

Bachelor's degree in computer science, engineering, or a related field, or equivalent practical experience; advanced degree a plus.
No certifications or licenses required.
15+ years in software and platform architecture or engineering, including senior engineering leadership (director level or above) or principal/consulting-architect scope across multiple teams.
Recent, hands-on depth in LLM-based and agentic systems carried to production — including agentic infrastructure itself: control planes, agent runtimes, gateways, SDKs, and the developer experience around them.
Founder or zero-to-one platform experience is highly valued.
Has personally designed and shipped an AI platform, agent framework, or ML platform at meaningful scale — and can walk through the decisions, trade-offs, and failures.
Experience operating as a principal or consulting architect across multiple teams — setting direction through standards, prototypes, and persuasion rather than authority alone.
Experience partnering with enterprise architecture, security, and governance functions in a large organization without losing product velocity.
Deep fluency in the modern agentic stack: orchestration frameworks, tool/function calling and MCP, retrieval (vector and graph), context and memory management, and evaluation methodology.
Strong data architecture background: lakehouse platforms, semantic and graph modeling, data contracts, and pipeline architecture for structured and unstructured data.
Knowledge representation: ontologies, knowledge graphs, GraphRAG, judgment/decision modeling.
LLM platform engineering: gateways, routing, evaluation harnesses, observability, token economics; model strategy across frontier and open models.
Enterprise integration: APIs, event-driven patterns, data contracts, identity and entitlements.
Writes production-quality code and prototypes; recent hands-on work in Python and/or TypeScript.
Demonstrates personal AI leverage: designs and prototypes with agentic tools as a force multiplier — the working standard for this team.
Builder-first temperament with strong architectural judgment; pragmatic about which decisions must be made once and well versus delegated; raises the technical bar of everyone in range; writes things down; makes complex systems explainable to executives and practitioners; treats the firm's platform, substrate, and runbooks as owned assets.

Responsibilities

Architecture design and standards. Own the target-state architecture for the AI product portfolio and the architecture decision records that make significant decisions transparent, durable, and teachable.
Design the agent platform: harness and orchestration patterns, tool and MCP integration, retrieval architecture spanning vector and graph, memory and context management, and the shared services every product consumes.
Co-design the intelligence substrate (ontology, knowledge graph, judgment layer) with the substrate lead.
Set and steward a model-agnostic model strategy — selection, routing, fallback, fine-tuning versus context strategies, and cost-per-outcome economics — and keep the platform portable and vendor-neutral.
Hands-on technical leadership. Design by building: prototype the patterns product teams then harden.
Review and shape each product's architecture with the engineering teams — hands-on, in the code.
Chair APG's technical design authority.
Lead architecture absorption from the firm's expert engineering partners on a shadow-pair-lead-own ramp, converting partner-built patterns into firm-owned standards, decision records, and runbooks, with defined gates at which each architecture layer becomes firm-owned.
Establish evaluation-first engineering as an architectural standard: no capability is production-ready until its evaluation suite says so.
Define the non-functional architecture (scalability, deployment, security, observability) with enterprise technology partners so products transition cleanly into enterprise operation.
Co-own the production handoff standard and exercise it on the first products through the gate.
Operate the architectural seam with Enterprise Architecture — full design authority inside APG products, built to enterprise standards, data contracts, and production gates.
Advise the CAITO on build, buy, and integrate decisions across the portfolio, including vendor AI architecture boundaries where the firm must own the data and knowledge layer.
Help hire and mentor the founding engineers; raise the bar on design quality, code quality, and written technical decision-making.
Represent APG architecture in the firm's architecture forums and, with the CAITO, to executive leadership.
Engage and participate in special projects or complete other duties as assigned.