Head of Intelligence Architecture

Why This Role Exists:
Rocket Equities is a boutique M&A advisory and asset management firm operating across six time zones. Every day, the firm generates intelligence through deal execution, investor conversations, market research, email correspondence, Slack discussions, and CRM activity. Today, that intelligence is scattered across tools and people’s heads. When someone leaves, the knowledge walks out the door.
This role exists to build the firm’s unified institutional brain — a single knowledge architecture that captures every deal pattern, investor preference, relationship signal, sector insight, and operational decision, making it queryable, actionable, and continuously self-improving.
We already use Cognee (knowledge graph with Neo4j + Qdrant), have MCP-based integrations across four email accounts, Slack, and calendar, and have ~155 investors mapped. You are not starting from zero. You are taking fragmented infrastructure and turning it into a production-grade intelligence system the entire firm depends on.
The Mandate:
Build the knowledge layer that makes the entire firm smarter. Every agent, every process, every decision should draw from a single, well-architected source of firm intelligence.
What You Will Own:
Knowledge Architecture
Design and maintain the firm’s ontology: entities (investors, companies, deals, sectors, people, funds), relationships, and events
Evolve the existing Cognee/Neo4j/Qdrant stack into a production-grade knowledge graph with proper schema governance
Build ingestion pipelines that automatically capture structured knowledge from emails (4 accounts), Slack, meeting notes, deal files, CRM, and Monday.com
Implement entity resolution so the system recognizes that “Andreia from KRIF” in an email and “Andreia Muresan” in a contact list are the same person
Retrieval & Intelligence Layer
Build RAG pipelines that let anyone in the firm query accumulated knowledge in natural language
Implement hybrid retrieval (vector + graph + keyword) so queries return contextually rich, relationship-aware answers
Create intelligence surfaces: deal pattern recognition, investor-sector affinity scoring, relationship strength indicators, communication gap alerts
Agent Foundation
Provide the knowledge API that all process automation agents consume — you build the brain, the automation team builds the hands
Ensure every agent built on top of your layer has access to full firm context, not just its narrow workflow data
Build feedback loops where agent outputs and user corrections flow back to improve the knowledge graph
Data Quality & Governance
Own data quality: deduplication, relationship validation, staleness detection, confidence scoring
Build monitoring that surfaces when knowledge is decaying (e.g., investor contact info is 6 months old, deal status hasn’t been updated)
Establish access controls — deal-sensitive information should only be queryable by authorized team members
Who You Are:
3-5 years of professional experience. You have built knowledge systems, search infrastructure, or data platforms that real organizations depend on in production. You think in graphs, entities, and relationships — not just vectors and embeddings.
Current title range: Senior AI/ML Engineer, Knowledge Engineer, Senior Data Engineer, AI Architect, Search/IR Engineer, Senior NLP Engineer.
You are not: a data scientist who builds dashboards, a DevOps engineer, a pure researcher, or a project manager. You are an engineer who architects and builds knowledge systems.
Required Skills
Must Have:
Python — production-grade, not notebooks
Agentic AI Engineering
Knowledge graphs — Neo4j or equivalent graph database, ontology design, entity resolution, relationship modeling
Vector databases — Qdrant, Pinecone, Weaviate, or ChromaDB in production
RAG architecture — hybrid retrieval (vector + graph + keyword), chunking strategies, re-ranking, citation grounding
LLM APIs — Claude and/or OpenAI for extraction, summarization, and query answering
Data pipeline engineering — building ingestion from multiple sources (email APIs, Slack, CRMs, documents) into structured knowledge
NLP fundamentals — named entity recognition, relation extraction, coreference resolution
Must be able to build interfaces that translates complex solutions into seamless user experiences
Stays current with the latest developments in AI and emerging technologies
Nice to Have:
MCP (Model Context Protocol) or similar agent-tool integration frameworks
Graph neural networks or graph-based reasoning
Experience with financial services data (deal records, investor CRMs, compliance documents)
LangChain / LlamaIndex / CrewAI for agent orchestration
Cloud deployment (AWS/GCP) and API development
Non-Negotiable Soft Skills:
Architectural thinking — you design systems that scale and evolve, not one-off scripts
English communication — you will present to the Managing Partner and explain knowledge architecture to non-technical deal professionals
Domain curiosity — you want to understand how M&A advisory and fund management actually work, not just treat it as “another data source”
Self-direction — you identify what knowledge the firm is losing and build systems to capture it, without being told
Compensation:
Base Salary: $3,000 USD per month
Performance Bonus: Tied to quarterly intelligence system milestones
Structure: 3-month contract-to-hire → Full-time
Working Hours: Flexible; overlap with CET/PHST time zones required
Equipment: Own laptop; all API costs, tool subscriptions, and cloud infra covered

Work arrangement
Yes