Open to senior & AI engineering roles

Gevorg Gasparyan
Full-Stack & AI Systems

6+ years shipping production platforms across CRM, health-tech, capital markets and e-commerce — on distributed teams spanning the UK, US, Denmark, Sweden and Canada. Today I design multi-agent AI systems that autonomously plan, reason and execute complex workflows on top of deterministic engines I can actually reason about.

Explore my platform → Download CV LinkedIn
◎ Yerevan, Armenia ◇ Remote-first ✉ gevorggasparyan33@gmail.com
Portrait of Gevorg Gasparyan
Currently @ Vaimo
6+Years shipping production
6Companies · 5 countries
48Domain modules built solo
2,000+Test suites written
Career

Where I've worked

Six roles, five countries, one constant: TypeScript end to end. Tap any position to expand the detail, stack and links.

Continuation of the Helphouse.io engineering team following its integration into Vaimo.

  • Design and build production multi-agent AI systems that autonomously plan, reason and execute complex customer-care workflows, using LangGraph and the Claude Agent SDK.
  • Architect LLM integrations, agent memory systems and tool orchestration, deploying scalable, secure services on cloud infrastructure.
  • Deliver bespoke AI integrations inside custom Zendesk applications, extending premium customer-care products well beyond out-of-the-box CRM capability.
  • Build tailored Zendesk Help Centers for enterprise clients, tightly coupled to each client's support model.
  • Own end-to-end delivery: system design, implementation, deployment, monitoring and ongoing optimization.
TypeScriptZendeskLangGraph Claude Agent SDKn8nReact Node.jsDjangoGCP

Client Help Centers shipped at Vaimo

★ Favorite client DBU — Dansk Boldspil-Union The Danish Football Association's official support Help Center — role-based flows for players, coaches, club leaders and team managers. The client I'm proudest to have built for. ↗ danskboldspilunion.zendesk.com Nortec Self-service Help Center for Nortec's laundry-payment and EV charge-point systems, covering users, landlords and administrators across multiple product lines. ↗ support.nortec.dk
↗ vaimo.com
  • Built and customized Zendesk CRM applications and Help Centers, coding client-specific functionality on top of the platform and ensuring clean integration with existing support workflows.
  • Owned post-implementation support alongside development — triaging client queries, diagnosing integration failures and resolving technical issues in live customer environments.
  • Worked within a distributed, multi-national team, continuously adapting to a fast-moving CRM and AI tooling landscape.
TypeScriptZendeskExpress.js ReactNode.js
↗ helphouse.io

Canadian health-tech company delivering accessible fertility care through technology.

  • Built backend services powering patient-facing iOS and Android apps plus a clinic-staff web portal.
  • Implemented appointment booking, doctor availability, schedule templates and test-result delivery.
  • Developed specimen-collection tracking and automated cryo-tank management for egg and sperm freezing — regulated, high-integrity clinical workflows where data correctness is non-negotiable.
  • Designed microservice boundaries and REST APIs with NestJS and TypeORM over MySQL; covered business logic with Jest.
NestJSTypeScriptNode.js MySQLTypeORMMicroservices REST APIsJest
↗ pollinfertility.com

Equity capital markets platform connecting banks and institutional investors.

  • Developed features publishing golden-source deal data directly from banks to institutional investors in digital format — the platform's core value proposition.
  • Built a customizable, real-time market-activity overview designed for interoperability with the banks' and investors' existing systems.
  • Contributed to ECM workflows engineered for regulatory compliance, accountability and auditability.
  • Worked across serverless AWS infrastructure — Lambda and DynamoDB — with Terraform-managed provisioning.
  • Recognized by colleagues for depth of code review and pair-programming contribution.
ReactReact QueryTypeScript TailwindAWS LambdaDynamoDB TerraformNode.js
↗ primaryportal.com

Online marketplace and white-label ordering / delivery platform.

  • Built and maintained both the admin panel and the white-label e-commerce platform powering product ordering and delivery.
  • Implemented and streamlined the ordering-and-delivery pipeline, keeping the experience navigable for customers and administrators alike.
  • Delivered GraphQL APIs backed by PostgreSQL, with Redis caching and RabbitMQ for asynchronous messaging.
TypeScriptNode.jsReact GraphQLPostgreSQLRedis RabbitMQ
↗ rebud.com ↗ hof.delivery

Innovation-management platform for crowdsourcing and evaluating ideas.

  • Built idea submission, voting and collaboration features across both front end and back end.
  • Implemented real-time collaboration with Socket.io, plus authentication and data-management layers.
  • Delivered UI on Material-UI and Ant Design component systems.
ReactTypeScriptNode.js Socket.ioMaterial-UIAnt Design
↗ ideascale.com
Personal Platform

OptiSchedule

An AI-native scheduling, booking and CRM platform I designed and built solo — the system where I work out what production-grade agentic software should actually look like.

AI-Native Scheduling & CRM

Solo build TypeScript monorepo

Users express intent in natural language; agents plan and optimize business operations on top of a deterministic scheduling and workflow engine. The whole system rests on one safety principle — AI never mutates state directly. Agents propose plans, the policy engine validates and risk-scores them, and only a deterministic executor is allowed to change anything.

Intent→ Context Builder→ Planner Agent→ Critic Agent→ Policy Engine→ DAG Compiler→ Deterministic Executor→ Event Store

◈ Execution-safety pipeline

The policy engine risk-scores every plan from low to critical, then routes it to autonomous, requires-approval or suggestion-only execution. Nothing reaches the database unvalidated.

◷ Deterministic scheduling engine

10-minute slot granularity, UTC storage with timezone conversion at the presentation layer, and double-booking prevented by pessimistic locking inside PostgreSQL transactions — modelling schedules, breaks, vacations, durations and buffers.

⛓ DAG workflow engine

Idempotent, retryable, fully audited steps with exponential-backoff retry policies and saga-pattern compensation on failure. Every execution history is persisted.

✦ LangGraph agent framework

Scheduling-optimization, cancellation-recovery, conflict-resolution, utilization-optimization and schedule-apply agents — plus a ReAct booking agent operating over a typed tool registry.

⌨ Natural-language command layer

Spans booking, CRM, catalog, payments, inventory, loyalty, marketing and clinical workflows — gated by a four-tier access-control matrix (client / staff / manager / owner) with per-data-category permissions.

▦ 48 domain modules

Bookings, memberships, subscriptions, gift cards, promo codes, retail POS, inventory, commissions, loyalty, referrals, reviews, analytics, billing, compliance, and a clinical suite (LIS, diagnostic codes, test results, questionnaires, intakes).

◱ Multi-surface delivery

A Next.js dashboard plus consumer and provider mobile apps built in React + Vite and packaged for iOS and Android via Capacitor, over an event-driven core with Socket.IO real-time updates.

✓ Engineering rigor

~2,000 test suites across the monorepo. Modular-monolith DDD boundaries designed for later service extraction, deployed on Vercel and Railway with Dockerized PostgreSQL 16 and Redis 7.

TypeScriptNestJSNext.js ReactLangGraphPostgreSQL 16 Redis 7TypeORMSocket.IO CapacitorDockerVercel Railway
Toolkit

What I build with

AI & Agents

LangGraphClaude Agent SDKClaude Code n8nGitHub CopilotTool orchestration Agent memorySemantic modeling

Languages

TypeScriptJavaScriptPythonSQL

Frontend

ReactNext.jsRedux React QueryTailwindMaterial-UI Ant DesignMicro-frontends

Backend

Node.jsNestJSExpress.js RESTGraphQLMicroservices Socket.IORabbitMQ

Data

PostgreSQLMySQLDynamoDB RedisTypeORM

Cloud & Infra

AWS LambdaGCPTerraform DockerVercelRailway GitJest
Get in touch

Let's build something that thinks

Open to senior full-stack and AI engineering roles, remote or hybrid. The fastest way to reach me is email.

gevorggasparyan33@gmail.com LinkedIn Download CV