Ahmed A.

I build production AI systems and the backend infrastructure they run on. On the AI layer: multi-tenant LLM gateways across Bedrock, OpenAI, and Azure, LangGraph agent orchestration with Postgres checkpointing, RAG on pgvector with OCR ingestion. On the layer below: banking integrations against payment aggregators, and SMS and USSD infrastructure at 5 to 10 million messages a day. Compliance-grade pipelines, travel and booking APIs, event-driven systems on Kafka and RabbitMQ. Three of these are live at ansatech.space: a multilingual RAG system, a document extraction pipeline, and a dispute resolution assistant. Working software you can click, not screenshots. Where I have real depth: Fintech and payments. Banking integrations against a national switch across multiple institutions, payment gateway work spanning four national payment rails, auditable ledgers where reconciliation has to survive a regulator. Research and compliance AI. Document ingestion with OCR and LLM extraction, meeting transcripts feeding a RAG pipeline, MCP integrations connecting AI assistants to internal tools with permission boundaries that hold. High-throughput messaging. I architected a multi-tenant SMS aggregation platform end to end, running on RabbitMQ and Kubernetes, plus USSD infrastructure for telco and banking clients. Very few engineers have done this at production scale. Travel. Flight and booking API integrations and partner onboarding into a live system. Stack: Python, FastAPI, TypeScript, NestJS, Java, Spring Boot, LangGraph, LangChain, PostgreSQL and pgvector, AWS (ECS, Fargate, RDS, Bedrock), Docker, Kubernetes, Kafka, RabbitMQ, SQS, Redis, Terraform. I am comfortable owning a system end to end, from the ingestion pipeline to the model routing to the thing that pages you at 3am. If that is the shape of your problem, tell me what is breaking and I will tell you honestly whether I am the right fit.