What I build
Backend systems, internal AI products, asynchronous processing, and production automation.
About
Production systems · AI products
Eduard Han
Backend and AI Product Engineer
I build production systems that replace expensive manual workflows.
From product discovery and architecture to deployment, evaluation, and production operation.
Selected evidence
Designed and built a platform processing 30,000+ automated reviews every month.
Core stack
Go, Python, Redis, SQL, Docker, Linux, ASR, and LLM integrations.
How I work
Prototype quickly, test with real data, observe production, and iterate from evidence.
Flagship project
Automated call quality control
Echo
I designed and built an internal AI platform that analyzes complete business calls and turns them into structured operational data.
Sole core product engineer: product design, architecture, backend, frontend, AI evaluation, infrastructure, and production operation.
Input Complete conversation Output Structured review
30,000+automated reviews
per month
per month
50+prompt-defined
review types
review types
~1 minuteend-to-end
processing
processing
100–200×lower estimated
operating costCompared with equivalent manual review capacity.
operating costCompared with equivalent manual review capacity.
Before
Selected call fragments · manual review forms · limited coverage · human training for every new check
After
Complete conversations · prompt-defined checks · structured results · near-complete operational coverage
Selected work
Systems beyond Echo
Different problems.
One operating standard.
Three projects selected for breadth: business AI, infrastructure automation, and a human-confirmed consumer workflow.
Multi-channel conversational sales platform
Relay
One delivery-aware core for WhatsApp, Telegram, and VK, with staged LLM dialogue, acknowledgements, guarded follow-ups, and multi-tenant funnels.
GoPythonPostgreSQLVue 3TypeScriptWhatsAppTelegramLLM gatewayDocker
Deal-context assistant with hybrid retrieval
Aileet
A manager-facing assistant that combines deal context from two databases, hybrid knowledge retrieval, validated structured output, and streamed responses.
GoPostgreSQLpgvectorMSSQLWebSocketJWT
Human-confirmed nutrition tracking with vision AI
Amazed By You
Food and training tracking where vision estimates meals, but only user-confirmed data reaches the diary.
GochiPostgreSQLJWTReactVitevision LLM
Contact
Available for the right problem
Need someone who can turn an unclear internal problem into a production system?
I’m open to backend, AI systems, internal product, and product-engineering roles.