AK
WorkExperienceCase StudiesTimelineWritingContact
Resume
AK

Athens, Greece

Now live: Matcha · my.bookmatcha.app

Aggelos Konstantilieris

Full-Stack Developer & AI Integrations Lead

I design and ship GenAI-powered product systems: AI agents, batch/realtime LLM workflows, prompt-action architectures, CRMs, and internal tools for teams that need reliable software. Co-founder of Matcha, an AI booking platform for WhatsApp.

View AI systemsStart a project

Building at LgMAR

Open to new projects

Experience

The teams and systems that shaped how I build.

I work best where product clarity, operational detail, and engineering rigor all matter at the same time.

Selected role

Matcha

Matcha

Co-Founder & Lead Developer

Remote - Athens, GR

May 2025 - Present

Co-founded Matcha (my.bookmatcha.app) and lead its engineering: a booking platform for appointments and classes where an AI assistant answers customer messages on WhatsApp and books once they confirm, a hosted booking page takes direct bookings, and one calendar keeps the team in sync. Live in early-access beta for fitness studios, salons and clinics, in Greek and English.

  • Architected the multi-tenant NestJS 11 backend with a 3-person founding team as a hexagonal, domain-driven monorepo — ~529K lines across 105 Nest modules with 263 ports and 273 adapters — alongside a Next.js tenant dashboard and operator console.
  • Built the LangGraph booking agent with 24 Zod-validated tools, three-tier input guardrails, and regex-first bilingual routing with deterministic shortcuts on ~40% of turns.
  • Made bookings race-safe with atomic MongoDB reservations, transactional outbox delivery, 38 scheduled jobs, Stripe billing and Google Calendar sync.
  • Shipped a ~15,900-case test suite and a ~270-scenario agent eval harness with release gates, so agent changes are promoted on measured behaviour rather than spot checks.
Visit my.bookmatcha.appRead the case study

Writing

Writing on reliable AI systems.

Notes from building production AI: evals, structured outputs, orchestration, and the operational layer around LLMs.

Featured article / 7 min read

Evaluation-Driven Development: Shipping an AI Booking Agent You Can Trust

Golden fixtures, simulated users, LLM-as-judge scoring, safety gates, latency budgets, and promotion guardrails for Matcha's AI booking agent.

Read the article
Browse all writing

Start a project

Building a GenAI workflow, AI agent, or operational product?

I help teams ship production AI systems — agents, LLM orchestration, and the CRMs and internal tools around them — as software they can trust every day. Matcha, the AI booking platform I co-founded, is live and taking bookings today.

Start a projectSee a case studyTry my.bookmatcha.app

Founder product · live in beta

Matcha — AI Booking Platform

Open case study

Customers book from the chat they already use while the business runs the day from one dashboard. Live in early-access beta at my.bookmatcha.app.

~40%
turns with deterministic shortcuts
~529K
lines of code
~15,900
test cases
Open case study

AI Systems & Product Work

AI systems and operational products built for real workflows.

Case studies across GenAI agents, LLM batch orchestration, prompt-action systems, CRMs, booking flows, and logistics platforms.

GenAI systems / batch orchestration / AI agents / operational software

Browse all case studies
Matcha — AI Booking Platform

Founder product · live in beta / AI Booking Platform

Live · my.bookmatcha.app

Matcha — AI Booking Platform

Live booking system for appointments and classes: an AI assistant answers customer messages on WhatsApp and books once they confirm, a hosted booking page takes direct bookings, and one calendar keeps the whole team in sync.

~40%
turns with deterministic shortcuts
~529K
lines of code
~15,900
test cases
Category
AI Booking Platform
Audience
Fitness studios, salons, clinics
Outcome
Customers book from the chat they already use while the business runs the day from one dashboard. Live in early-access beta at my.bookmatcha.app.

Built with NestJS / Next.js / OpenAI / Redis / TypeScript

View case study
Forge

Latest AI infrastructure project / LLM Infrastructure

Forge

Provider-agnostic system for running LLM batch and realtime jobs across OpenAI and Anthropic, with automatic retries and crash recovery.

~40/s
jobs per replica
~50%
lower token cost
2
LLM providers
Category
LLM Infrastructure
Audience
Teams running batch and realtime AI jobs
Outcome
Reliable, scalable AI job processing that keeps running through crashes and provider rate limits, without needing a separate queue service.

Built with Node.js / TypeScript / PostgreSQL / OpenAI / Anthropic / Prometheus

View case study
Omni AI Integration Architecture

Professional AI integration work / AI Integration Architecture

Omni AI Integration Architecture

AI-augmented maritime CRM architecture for classification, entity extraction, prompt-action workflows, and operational automation.

~3,000
platform users
4.5★
mobile app rating
88%
AI-pipeline bugs landed
Category
AI Integration Architecture
Audience
Maritime operations and communication teams
Outcome
A modular AI integration foundation for extracting operational meaning from communication workflows.

Built with React / TypeScript / Node.js / OpenAI

View case study
Apollon CRM

Featured case study / Operations CRM

Apollon CRM

Full-stack CRM for pet boarding teams with room scheduling, client records, booking management, and financial reporting.

~€794K
transactions tracked
784
clients
73
boarding rooms
Category
Operations CRM
Audience
Pet boarding and care businesses
Outcome
One workflow for bookings, occupancy, pets, owners, and revenue.

Built with React / TypeScript / Tailwind CSS / Next.js

View case study

Additional work

Fotokiklosi

Additional work / Logistics Platform

Fotokiklosi

End-to-end waste management platform coordinating collection points, routes, material handling, and internal reporting across Greece.

Audience
Waste management operations teams
Outcome
Route planning, traceability, and internal workflows in one production system.

Built with React / TypeScript

View case study