Andrii Vasyliev
Contact

Web Platforms · AI Systems · Robotics R&D

Andrii Vasyliev — Engineering reliable systems, automation & AI

I spent over a decade working on the Brand Management System platform at MetaDesign — a multi-tenant web platform used by global brands. My core craft is web and platform engineering; today I extend that mindset to AI agents, automation, and domain-specific languages.

Day to day, most of my work is still full-stack web platforms — backend, frontend, and the infrastructure in between.

Robotics prototypes and autonomy rigs are my long-term personal R&D track, shaping how I think about AI systems under real-world physical constraints. I’m not actively looking for a full-time role right now, but I’m open to collaborations around platform architecture, AI automation, or robotics research.

Contact me View projects LinkedIn
WebAI agentsPHPPythonLangGraphLangChainPostgresDockerTraefikVue 3DDD/SOLID
Portrait of Andrii Vasyliev

12+

years of software & systems engineering, mainly web platforms

Loops

observability and test feedback built into every launch I work on

AI × OT

workflows that connect agents, operators, and robotic systems

Readiness

from discovery to launch — fast, measurable, reliable

Systems expertise

Building products that survive the real world

I treat a product as a single system — starting from long-lived web platforms and their data flows, and extending into AI agents and robotics R&D. The goal is to bring the reliability of enterprise web systems into automation loops and experiments that reflect real-world constraints.

Background

Most of my career was at MetaDesign in Berlin, where I spent 12+ years evolving the Brand Management System web platform for global clients — architecture, performance, and long-term stability. That base in web and platform engineering now anchors my work as Chief Research & Development at Funely AI, where we design agent-driven automation for service businesses. In parallel I run personal robotics R&D — chassis, sensing, and control rigs that inform how I design systems before they reach production environments.

Methodology

  • Discovery → prototyping → production with clear checkpoints and rollback paths
  • DDD/SOLID as the backbone for dependable architecture
  • Embedded test scenarios and basic telemetry instrumentation by default

Outcomes

  • Transparent handover from prototype to day-to-day operations
  • Agents and humans working in a single, observable loop
  • Readiness for fast iteration without losing control of quality

Core focus areas

Robotics

Personal R&D on chassis, kinematics, sensing, and drive systems through prototypes and field-like trials.

AI + DSL

Designing domain languages with AI copilots that automate and monitor real operational workflows.

Production engineering

Working with Docker/Traefik setups and extending existing CI/CD and observability so releases stay predictable at small scale.

Robotics

Experimental off-road robotics for sustained load

Personal R&D across chassis and systems aimed at working through mud, dust, and continuous schedules. Kinematics, power, and sensing modules are designed as prototypes to explore how platforms could handle repeated field-like runs with heavy payloads and off-road speeds.

Personal R&D

Mechanics

Suspension modelling, custom joints, and balance studies to keep experimental platforms stable under shifting loads.

  • Tracked and wheeled concept platforms
  • Modular sensing stations and mounts
  • Dynamic skew and load compensation experiments

Electronics & control

Working with BLDC drives and onboard controllers to capture telemetry and explore how agents could sync with field scenarios.

  • CAN bus integration, basic power delivery and protection concepts
  • Planning around IMU, GNSS, LiDAR, and video pipelines for perception
  • Early designs for safety loops and emergency control paths

Operational readiness

Building simple rigs, test tracks, and service ideas to support quick upgrades and repairs on experimental platforms.

  • Basic simulation setups for load and wear behaviour
  • Telemetry pipelines for logging runs and experiments
  • Notes and procedures that can evolve into maintenance playbooks

DSL-Hub

AI-assisted domain languages — in active development

DSL-Hub is my ongoing project exploring how AI agents can help design, validate, and run domain-specific workflows. The architecture combines a structured authoring layer with a controlled runtime, while keeping the whole system auditable and predictable.

Authoring Plane

  • AI copilots draft early flow specifications from process briefs
  • Schemas & validators maintain compatibility during iteration
  • Migration experiments include AI-assisted risk checks

Runtime Plane

  • Deterministic execution graphs (early prototype)
  • Event logs & traces for debugging and observability
  • Safe rollback experiments for iterative releases

What I’m solving

  • Keeping multiple DSL versions consistent during rapid iteration
  • Using AI to reduce migration and validation errors
  • Making automation transparent through clear execution traces
  • Translating real-world processes into governed pipelines
  • Letting humans and agents co-author logic with guardrails

AI Startups

Agent-driven workflows for fast iteration

At Funely AI and similar early-stage projects I build lightweight frameworks for rapid LLM and agent experimentation — where ideas can move from prototype to internal production extremely quickly.

Product workflows

  • Prompt-linked state machines with versioned logic
  • Agent integrations with back office, CRM, and billing APIs
  • Analytics loops and automated response evaluation
  • Browser and Chrome-extension tooling for operators and internal teams

Operational maturity

  • Adapting CI/CD and infrastructure-as-code setups for fast experiments
  • Safe rollout and rollback patterns for iterative releases
  • Prototyping multimodal and multi-agent workflows

Selected Work

Projects where I owned the full lifecycle

From early technical discovery through launch and operations, these projects demonstrate how I combine AI, systems engineering, and delivery.

DSL-Hub

Platform

Authoring, migration & runtime guardrails for operational DSLs

A platform currently in active development that combines DSL generation, automated validation, and observability so teams can align logic fast without losing control at release time.

Differentiators

  • AI-assisted generation with human-in-the-loop audit
  • Event-sourced runtime
  • End-to-end execution tracing

Focus

  • Reduce time to update DSL logic
  • Make migrations safer and less disruptive
  • Give product teams clear, DSL-level metrics

Brand Management System (MetaDesign)

Enterprise

Long-term web platform for global brand management

Multi-tenant platform that keeps brand assets, guidelines, and tooling consistent for international teams. I was responsible for architecture evolution, stability, and integrations for over a decade of production use.

Focus

  • Multi-tenant architecture and governance
  • High-availability brand and asset management
  • Enterprise integrations and APIs

Results

  • Stable delivery for global clients over many years
  • Controlled feature rollouts for distributed brand teams
  • Consistent digital assets across markets and channels

Funely AI

Product

AI-driven customer acquisition and operations for appliance repair and service businesses

Funely AI connects lead generation, landing pages, and automation into one pipeline. I lead the technical architecture and CR&D (Chief Research & Development), translating messy real-world workflows into repeatable, measurable flows.

Focus

  • AI-assisted lead funnels and intake flows
  • Connected CRM, billing, and scheduling
  • Fast A/B testing loops on real traffic

Results

  • Clear attribution from click to booked job
  • Faster iteration on campaigns without chaos
  • A technical base ready to scale to new verticals

Tech Stack

Full tech & engineering stack — Andrii Vasyliev

Battle-tested web and platform foundations, deep domain experience, and modern AI-driven tooling grouped by how I apply them in real programmes.

AI & Agents

Agent-driven automation loops and domain-specific copilots.

LLM orchestrationMulti-agent orchestrationPrompt-linked DSLsTelemetry feedback loopsAutomated QA loops (R&D)RAG + tool integrations

Backend & Data

Web platforms and data flows built to stay measurable for years.

FastAPI / Python servicesNode.js / TypeScript runtimesPHPLaravelSymfonyYiiPostgreSQLMySQL / MariaDBRedisRabbitMQEvent-driven architectures

Ops & Delivery

Containerized services, pragmatic observability, and fast iteration.

Docker & ComposeLinuxNginxApache2Traefik v3K3s / Kubernetes experiments

Frontend & UX

Interfaces and tooling that support operators and internal teams.

Vue 3 + ViteTypeScriptTailwind CSSPrimeVue systemsComponent architectureChrome Extensions

Contact

Tell me about the challenge — I reply personally

For web platforms, AI automation, DSL-Hub, or robotics R&D, leave a short context. Messages are handled through a private backend and never shared with third parties.