Job Description
professionals an intelligence advantage over the market. Underneath sits a structured data platform on Snowflake and AWS
covering commercial property transactions, tenant occupancy, availability, investor activity and financial KPIs — with a semantic
layer that lets agents query this data through natural language. The candidate will design and ship multi-agent workflows end to
end: orchestrating LLM tool-calling across data sources, building retrieval and generation pipelines for document intelligence, and
connecting agents to the surfaces where brokers work — from lease expiry targeting and heads of terms review to client pitch
generation, investment matching and market monitoring. The role owns the full lifecycle from scoping use cases with business
SMEs through to production deployment on cloud infrastructure in wave-sequenced sprints.
Must-have Skills
LLM Agent Development — designing multi-agent systems
with tool-calling, chain-of-thought reasoning, structured outputs
and context management using Claude, OpenAI or Azure
OpenAI
Agentic Orchestration — building agent graphs and pipelines
(LangGraph, LangChain or equivalent); routing, disambiguation,
fallback handling and multi-step task decomposition
Python (5+ yrs) — async, Pydantic, FastAPI; building tool
interfaces, API services and data pipelines for production agent
systems
RAG & Document Intelligence — retrieval-augmented
generation, document extraction, summarisation and content
generation for business cases, client decks and reports
Cloud & Data Platforms — AWS (S3, Lambda, Bedrock,
SageMaker, Glue, Step Functions) and/or Snowflake; deploying
and scaling agent workloads on cloud infrastructure
SQL & Semantic Modelling — complex joins, CTEs, star
schemas; designing semantic layers that ground LLM agents in
structured data with verified queries and NL-to-SQL pipelines
Agent Evaluation & Guardrails — accuracy benchmarking,
hallucination detection, output validation, human-in-the-loop
review patterns and safety guardrails
Enterprise Integration — connecting agents to CRM,
SharePoint, Teams, email and document management systems
where business users work
NICE TO HAVE
Commercial real estate or professional services domain
knowledge
AWS Bedrock / SageMaker for LLM hosting and fine-tuning
MCP (Model Context Protocol) server design
Snowflake Cortex Analyst / semantic views
RAG pipeline design (Pinecone, Weaviate, FAISS)
Agent observability (LangSmith, Arize, Phoenix)
AWS CDK / CloudFormation for infrastructure-as-code
Agentic UX design for non-technical users
TECH STACK
Claude API OpenAI API LangGraph LangChain AWS
Snowflake Python FastAPI Pydantic Bedrock
SageMaker S3 Lambda Glue MCP Protocol Cortex
SQL Pinecone Weaviate RAGAS LangSmith Docker
K8s GitHub Actions
Interested in building AI for UK Advisory
Candidates should submit a CV, brief cover note, and links to any LLM agents, agentic workflows, or demos they've built
