Hi, I'm
Sachin Philip Mathew!
I build AI systems that work in production.
13+ years shipping high-impact products—from big data platforms to LLM-powered agents. Currently building the first AI RealEstate Agent at Everreal, where I lead architecture for a €1M+ ARR platform handling millions of daily transactions.
I specialize in the hard parts—context compression, provider failover, and making LLMs remember what matters without blowing up your token bill.
Staff Software Engineer
@ Everreal • 2020 - Present
Lead Product Engineer
@ Amura • 2018 - 2020
Senior Product Engineer
@ FullContact • 2016 - 2018
Product Engineer
@ Profoundis • 2012 - 2016
AI & LLM
LangChain • MCP • RAG
HuggingFace • Ollama
AWS Bedrock • Groq
Backend & Frontend
Python • FastAPI • Django
Node.js • TypeScript • Express
React • Next.js • Remix
Infrastructure
AWS • Terraform • Docker
PostgreSQL • OpenSearch • Redis
Building Cost-Effective, Production-Ready AI Conversational Assistants
How to build AI that remembers, recovers from outages, and costs 70% less—without blowing up your token bill.
München
Bavaria, Germany
Current Time
CET (UTC+1)
Staff Software Engineer at Everreal GmbH
Technical lead in 3-person team, co-creating roadmap for €1M+ ARR SaaS. Building the first AI RealEstate Agent. Implemented internal MCP to unify data sources. Led migration reducing API times from seconds to sub-100ms. Designed system handling 1M automated transactions daily.
AI Real Estate Agent
Building an AI agent that handles tenant inquiries, schedules viewings, and answers property questions—without human intervention. The challenge isn't making it smart. It's making it reliable, affordable, and able to recover when things break.
This means implementing MCP for real-time property data, context compression to keep conversations coherent without token bloat, and failover systems that switch providers mid-conversation if one goes down.