LangGraph · Distributed Systems · Financial Transaction Platform Full-time · Senior level
About the Role We’re looking for a senior Python engineer to help evolve and scale a critical transaction processing platform that handles millions of financial transactions every month. This is not a CRUD application, a chatbot project, or an AI prototype.
You’ll be working on a distributed, event-driven system responsible for ingesting, processing, classifying, enriching and synchronising financial transaction data across multiple services and products. The platform sits at the centre of our ecosystem and communicates with systems written in Python, PHP and TypeScript.
The role requires someone who is equally comfortable designing distributed systems, building production-grade APIs, optimising LLM-powered workflows, and understanding the consequences of architectural decisions across multiple interconnected services. Experience with LangGraph, AWS SQS, SQL and LLM optimisation is mandatory.
A significant portion of the platform relies on graph-based workflows for transaction classification, orchestration, enrichment and decision-making. We are looking for engineers who have built and operated LangGraph systems in production, understand event-driven architectures at scale, and know how to optimise LLM usage through intelligent caching, batching, routing and cost controls. Every change has a ripple effect across queues, APIs, databases, classification pipelines, tax calculations and customer-facing products. Success in this role requires strong systems thinking, attention to detail, and the ability to anticipate downstream impacts before they become production incidents.
The Platform Our transaction processing platform is responsible for:
Processing millions of financial transactions
Parsing structured and unstructured financial data
Classifying transactions using LLM-powered workflows
Extracting and enriching financial information
Synchronising data across multiple internal services
Feeding downstream bookkeeping, tax and reporting systems
Operating asynchronously through distributed worker architectures The platform combines traditional backend engineering, distributed systems design and modern AI orchestration techniques.
Technology Stack
Python 3.13 FastAPI SQLAlchemy
MySQL LangGraph AWS SQS
Event-driven workers
Asynchronous processing
pipelines
Pandas
PDF processing
LangFuse
Pytest
What You’ll Be Doing
Designing and implementing new transaction processing features
Building and maintaining LangGraph workflows
Improving transaction classification accuracy and reliability
Designing robust event-driven processing pipelines
Optimising LLM utilisation, latency and cost
Implementing intelligent caching and routing strategies
Building APIs consumed by PHP and TypeScript applications
Designing integrations between multiple internal services
Contributing to architectural decisions across the platform
Improving throughput, reliability and operational efficiency
This role involves significant ownership and influence over technical direction. Example Problems You Might Solve 1. Designing a classification workflow capable of processing hundreds of thousands of transactions daily without creating bottlenecks. 2. Building cache strategies that reduce LLM costs by 70% while maintaining classification accuracy. 3. Designing idempotent processing pipelines that safely recover from worker failures and duplicate events. 4. Tracing inconsistencies across Python, PHP and TypeScript systems to identify the root cause of financial data discrepancies. 5. Designing queue architectures that remain performant as transaction volumes increase significantly. 6. Identifying and mitigating subtle downstream impacts of schema, workflow or classification changes before deployment.
Mandatory Experience !!! Please do not apply unless you have experience with all of the following !!!:
Strong Python backend engineering experience
Prod
