system.identity
I build software that stays
reliable, maintainable, and scalable as complexity grows.
Over the past six years, I've designed, built, and operated production software from database design to deployment. I take ownership across the entire software lifecycle, focusing on backend architecture, business-critical platforms, and practical AI features that solve real business problems.
Tehran, Iran · Open to remote opportunities in Europe · Open to relocation
Hadi Bakhshi
Turning complex business problems into maintainable software.
What I optimize for:
• Long-term maintainability over short-term convenience
• Products that solve real business problems
• AI features people actually use
• Systems that make future development easier
system.impact
6+ Years
Engineering production software
2,600+ Network
Platform reach across dealership operations
~600 Callcenter Operators
Daily support via production RAG workflows
9 Modules
Unified in an internal ERP-style platform
End-to-End
From architecture and data design to deployment
system.about
Engineering approach.
The engineering problems that interest me most emerge after software reaches production. That often means defining service boundaries, designing data models, improving deployment workflows, or simplifying complex business processes.
My work typically spans the entire software lifecycle, from architecture and implementation to production operations and continuous improvement. I build systems that make future development easier for both the product and the team. More recently, that has included bringing AI into production through RAG pipelines and LLM-powered features designed with the same focus on reliability, observability, and operational resilience as any other production system.
system.work
Selected projects.
Reception Assistant
An internal platform that helps dealership receptionists prepare for customer interactions using historical survey data, behavioral prediction, and AI-assisted insight — turning millions of call-center survey responses into decisions a receptionist can act on in under a minute.
100 agencies · 2,600+ dealership network
Reception Assistant ETL
A real-time, event-driven pipeline built this year to replace three years of manual data uploads — syncing call-center survey data into Reception Assistant through RabbitMQ topic exchange, inbox/outbox messaging, and a configurable survey-mapping engine.
Millions of records · manual upload → real-time sync
RAG FAQ System
A production retrieval system for call-center operators that answered routine internal questions instantly and routed anything outside the knowledge base to the right department automatically.
600 daily operators · ~1 workday saved per person each week
Razeman
An ERP-style internal platform unifying HR, finance, and operations into one modular system, replacing a fragmented landscape of disconnected tools across a holding company.
9 modules · thousands of users
system.archive
More projects.
Reception Assistant ETL
Event-driven data pipeline with a runtime management API — operators configure field mappings and control pipeline behavior without redeployments.
Survey Nexus
A framework-agnostic survey engine supporting question logic, conditions, quotas, and SEC classification in a Turborepo monorepo.
Retail Audit
Retail audit management app — product/barcode audits, configurable questions, and detailed reporting for admins and auditors.
CAPI
Field survey app with real-time voice recording, location tracking, and quota-based task assignment for interviewers and admins.
Sales Service Inspection Dashboard
Automotive inspection insights dashboard comparing dealership performance, selling strategy, and CSI scores across the network.
Sales Service Inspection
Modernized after-sales inspection workflow app — redesigned UI, refactored front-end/back-end, and a renewed database design.
Motorcycle & Machinery Inspection Dashboards
Dashboards presenting years of after-sales inspection data through dynamic ECharts visualizations.
After Sales Service Inspection
Dynamic checklist-based vehicle inspection app with scheduling, automated PDF output, and reporting for the automotive industry.
system.stack
What it's built with.
Languages
Backend
Frontend
Databases
AI & LLM
DevOps
Architecture
Testing
system.experience
Where it happened.
- Owned architecture, backend, database design, and frontend of Reception Assistant — an AI-powered dealership platform used by 100+ agencies across a 2,600+ network. Led a major upgrade with LLM text generation, redesigned UI/UX, bulk analysis, and self-trained model integration.
- Built a production RAG-based FAQ system using n8n, pgvector, and OpenAI, serving ~600 call center operators daily and saving roughly a full workday per person per week.
- Owned full-cycle design and development of Razeman, an ERP-style integration platform unifying 9 organizational modules for thousands of users across a holding company.
- Designed and built Reception Assistant ETL — an event-driven pipeline with a runtime management API for field mappings and pipeline control without redeployments.
- Contributed to Follow-Up ETL using ASP.NET Core, RabbitMQ, and SQL Server, reducing data latency from over a week to near real time for 3,000+ users.
- Contributed to Survey Nexus, a framework-agnostic survey engine in a Turborepo monorepo, now powering CAPI and Retail Audit applications.
- Owned DevOps across production, staging, and dev — GitLab CI/CD pipelines deploying to Docker Swarm via Nexus, server infrastructure, and production incident response.
- Drove performance improvements through SQL indexing, query rewrites, caching strategies, and architectural refactoring. Mentored 3–4 engineers.
- Owned ASI and SI inspection applications for ISQI — inspector workflows, barcode validation, configurable question sets, and automated report generation with Express.js, SQL Server, and React/Next.js.
- Owned ASI and SI inspection dashboards — multi-role reporting platforms for dealership performance benchmarking and CSI score review, with admin panels for user and report management.
- Contributed to motorcycle and machinery after-sales inspection dashboards with dynamic ECharts visualizations, built with React, Vite, and Redux Toolkit Query.
- Optimized SQL Server and PostgreSQL queries, redesigned schemas for evolving requirements, and managed deployments on Linux with Nginx and SSL. Championed TypeScript adoption.
- Developed responsive React/TypeScript UI features, contributing to a 10% rise in user engagement metrics.
- Built shared component libraries that increased development velocity by 15% across multiple projects.
- Improved page load times by 1.5 seconds through code splitting, lazy loading, and bundle optimization.
- Integrated frontend features with REST APIs and improved state management using Redux and Context API.
system.profile
Background.
system.education
B.A. in English Translation
University of Applied Science and Technology, Tehran
Oct 2020 — Aug 2023
system.languages
- English — C1 — Professional working proficiency
- German — B1 — Actively learning
- Persian — Native