Quentin Mortier
AI Platform Engineer - Regadas, Portugal - EU work authorization
AI Platform Engineer, 13+ years across fintech and property tech. At GuestReady I own the platform end to end: AWS infrastructure, service architecture, application code, and for the last two years the AI layer, including a messaging agent that handles 80% of guest conversations in production.
I am looking for a role where I own the AI platform: taking agents from demo to something a business can safely run on.
Experience
AI & agents: Lead Architect, GuestReady
- Remote
Software architect for the whole platform, working across a 30-developer engineering team. I own the technical direction end to end, from AWS infrastructure and service architecture to the application code and the AI layer, and I still ship code every week.
AI & agentic systems
- Built Maia, our production guest-messaging agent (LangGraph, LangSmith). It handles ~80% of guest messages, around 3,000 a day, taking real actions like reservation changes and inquiry approvals.
- Brought guest first-response time down from 20 minutes to 2, and guest satisfaction on communication up 0.2 points (5-point scale).
- Hardened Maia with jailbreak pre-checks, answer verification and human handoff, measured with an LLM-as-judge eval harness.
- Designed and built a multi-tenant platform where users create their own agents, either through a no-code builder or by describing them in plain text. 40 active clients and ~400 conversations a day so far, mostly quality checks (photo analysis of cleaning) and incident notifications to Slack.
- Migrated half of the company's external automation flows onto the platform (target: 80%). Every agent action passes through a policy engine (auto, dry-run, escalate or deny) with budgets, circuit breakers and an audit trail, over a registry of 45+ tools on the company API.
- Added multi-agent delegation, human-in-the-loop runs that pause for approval and resume, and offline replay to test candidate agents against real event history without side effects.
- MCP in both directions: an OAuth 2.1 server exposing the company API to Claude and other AI clients (300 requests/day and growing), and a client layer letting agents use third-party MCP tools.
Platform & infra: Lead Architect, GuestReady
- Remote
Software architect for the whole platform, working across a 30-developer engineering team. I own the technical direction end to end, from AWS infrastructure and service architecture to the application code and the AI layer, and I still ship code every week.
Architecture & infrastructure
- Led the break-up of a monolithic application into microservices, using domain-driven design to draw the service boundaries and RabbitMQ events for inter-service communication, so teams could ship independently.
- Moved the platform from Heroku to Kubernetes on AWS (EKS): ~45 containerized services re-platformed with near-zero downtime. Infrastructure costs dropped ~40%, and outages went from a dozen a year to a single one.
- Built the channel manager integration with Airbnb, Vrbo and Booking.com (REST and GraphQL APIs), keeping listings, availability and reservations in sync for ~10,000 properties, around 300,000 sync events a day.
- Owned the PostgreSQL fleet (14 RDS instances): performance, availability and upgrades, including a rework of the notification pipeline that cut overall database load by 40%.
Team & practices: Lead Architect, GuestReady
- Remote
Software architect for the whole platform, working across a 30-developer engineering team. I own the technical direction end to end, from AWS infrastructure and service architecture to the application code and the AI layer, and I still ship code every week.
Team & engineering practices
- Built the company's internal Claude Code plugin marketplace and skills. Half the engineering team (19 of 39 active committers) now ships Claude-assisted commits in a given month.
- The review skill cut typical PR review time from ~20 minutes to ~5, mostly by generating an architecture diagram of each change so reviewers understand it at a glance.
- Moved team documentation to OKF, an agent-readable format, with automatic syncing: docs get updated in the same PR as the code they describe, so they no longer drift.
- Transverse role across a ~30-engineer team: defined and coached the practices for DDD, observability, testing and CI/CD.
Software Engineer: Software Engineer, GuestReady
- Remote
Full-stack engineer on the core product (Python/Django, ReactJS), from new modules to production performance.
- Developed core product modules, including the pricing engine, which reprices the whole 10,000-property portfolio every day plus ~2,000 reservation-triggered updates, and the guest messaging system (~500 messages a day at the time).
- Designed and built the company's first partner-facing REST API (Django REST Framework). It became the foundation for the channel manager integrations and now has 130 connected applications.
- Replaced 3 hours a day of manual cleaning-schedule planning with a constraint solver (Google OR-Tools) and a React review interface: the daily schedules for the in-house cleaning teams now take 10 minutes to generate and review.
- Tracked down performance issues across the stack with APM, log analysis and database profiling, from slow queries to architecture-level fixes.
Murex: Performance Stream Leader, Murex
- Paris
Four years at Murex, the capital-markets software vendor (MX.3), first in customer support, then leading the performance stream.
Performance stream leader (2 years)
- Led the performance stream across client accounts: KPIs, benchmarks, release deadlines, coordination between customers and Murex teams.
- Wrote the performance tests used to validate each release at production volumes.
- Fixed performance bottlenecks in C/C++ (Valgrind, Quantify, Purify), Java, and SQL on Sybase and Oracle.
Customer support engineer (2 years)
- Supported the production MX.3 environments of a large portfolio of client banks.
- Resolved critical issues fast, by whatever means worked: C/C++ and Java debugging, hotfixes, database indexes, on-site when needed.
EPITA: Master, MS in Computer Science, EPITA
- Paris
- Master, MS in Computer Science (Diplôme d'Ingénieur).
Toolbox: Areas of expertise, Toolbox
AI & Agents
- LangGraph
- LangSmith
- MCP
- RAG
Cloud
- AWS
- Kubernetes
- Docker
Programming
- Python
- Django
- FastAPI
- React
- PostgreSQL
CI/CD
- GitHub Actions
- AWS CodePipeline
- CircleCI
Event-Driven Architecture
- Domain Driven Design
- RabbitMQ
Off the clock: Languages & hobbies, Off the clock
Languages
- French: native speaker
- English: highly proficient
- Portuguese: highly proficient
Hobbies
- Lifting weights, then baking cakes.