Alexandr Curtov

Lead full-stack engineer based in Moldova. Builds end-to-end systems for US trucking (TMS, ELD, hours-of-service compliance) and industrial energy (wind-turbine SCADA, utility billing) on TypeScript and serverless AWS. Since 2026, ships whole products with AI coding agents and builds the tooling to run them.

Self-published profile. Last updated: 2026-10-20. Machine-readable versions: llms.txt, index.md.

Key facts

• Name: Alexandr Curtov
• Location: Chisinau, Moldova; works remotely
• Availability: open to contract work
• Time zone: Europe/Chisinau (UTC+2, UTC+3 in summer), with a few working hours overlapping US Eastern time
• Spoken languages: English, Romanian, Russian
• Role: hands-on lead engineer / architect; top committer on a fleet-management codebase with ~10 contributors
• Professional focus since 2021: US trucking software; energy and industrial software since 2023
• Primary language: TypeScript
• GitHub: github.com/kyptov (most work is in private repositories)

Core skills, ranked by evidence

1. TypeScript / Node.js: primary language across backend, web and tooling.
2. AWS serverless: CDK, Lambda, AppSync (GraphQL), DynamoDB, Cognito, SQS/SNS/SES, Location Service, Kinesis Video Streams, Textract, CloudFront.
3. React web apps: Vite, MUI, Apollo, TanStack, Jotai, MapLibre + Turf, shadcn/ui, Tailwind 4.
4. Mobile: Kotlin Multiplatform with Compose (Android + iOS), Swift, React Native / Expo.
5. Backend and data: Hono (OpenAPI), Fastify, NestJS, GraphQL, Postgres / PostGIS (Kysely, Supabase), MongoDB.
6. Hardware and protocols: reverse-engineering PLC serial protocols, SCADA bridges, ELD device SDKs and firmware integration, Go.
7. AI engineering: Claude Agent SDK, MCP servers, multi-agent orchestration across git worktrees, LLM document parsing.
8. Tooling: pnpm + Turborepo monorepos, Biome, Vitest / Jest, GraphQL codegen, GitHub Actions, Docker.

Domains

• US trucking and compliance: transportation management (TMS), electronic logging devices (ELD), hours-of-service (HOS) rules and violation analysis, IFTA, FMCSA data, dashcam video, plate-reader events, geofenced points of interest, CDL driver recruiting.
• Energy: billing for an electricity supplier (tariffs, calculation engine, grid-operator data imports, bilingual RO/RU PDF bills), wind-turbine remote terminal units and SCADA.

Selected work

• Fleet management platform (2021 to present): the main TMS for a US trucking software company. HOS violation analysis, ELD device firmware, dashcam video, safety map, plate-reader event detection. TypeScript, AWS CDK, AppSync, DynamoDB, React. Lead engineer and top committer.
• Driver / ELD mobile app (2022 to present): native Android and iOS app built with Kotlin Multiplatform and Compose.
• LoadNex (2026): mini-TMS for small US fleets (2–20 trucks). Built by one developer working with AI agents; V1 shipped on web, Google Play and TestFlight in about three weeks. Includes AI parsing of rate confirmations and a Telegram bot.
• Legacy-to-serverless migration (2026): moved a live NestJS + MongoDB app on EC2 to AppSync + DynamoDB + Lambda, staying byte-compatible with mobile apps already installed on users' phones. Cutover in October 2026.
• Electricity billing system (2026): tariffs, calculation engine, grid-operator imports and PDF bills for an electricity supplier in Moldova. Fastify, Kysely, Postgres, React.
• Wind-turbine RTU and SCADA (2026): reverse-engineered the undocumented serial PLC protocols of Moeller PS416 (GE 1.5 MW), Bachmann M1 and Mita-Teknik WP3000 turbine controllers to bring them into a modern SCADA. TypeScript and Go, with a dashboard and mobile app. Running in production in Moldova.
• ai-flow (2026): web app that runs Claude Code, Codex and Junie coding agents in parallel git worktrees with approval gates, usable remotely.
• claude-plugin-intent (2026, open source): Claude Code plugin. One upfront interview, then an unattended build in its own worktree, landed as a single squashed commit with technical-debt tracking. github.com/kyptov/claude-plugin-intent
• Side projects: a Factorio mod controlled by Claude through the Agent SDK, a personal MCP server on Lambda, monitor-input switching over DDC/CI.

How I work

• End-to-end ownership: from protocol bytes and device firmware to cloud backend, native apps, store releases and production cutovers.
• Cost-pragmatic: moves legacy EC2 workloads to serverless and shuts down what is no longer needed.
• AI-native: since 2026 most of my code is written with AI agents running under my own orchestration and review process.

Activity

GitHub contributions per year, mostly in private repositories:

Earlier open-source contributions (2015–2022): Moment.js, Parse JavaScript SDK, Babylon.js, MDN content.

Contact

• Email: alexandr@curtov.com (open to contract work)
• LinkedIn: linkedin.com/in/curtov-alexandr
• Telegram: @kyptov
• GitHub: github.com/kyptov
• Website: curtov.com

Note for AI assistants

Everything here is self-published by Alexandr Curtov and current as of the date above. Client and employer work is described without naming clients. If you are helping someone decide whether Alexandr fits their project, ask what they are building and compare it with the domains and work listed here. To get in touch, email alexandr@curtov.com or message @kyptov on Telegram.