Abdullah Gökmen — AI engineer

Republic of Türkiye Ministry of Health Ankara, Türkiye

question editable

pipeline

Entity extraction1 entity · 1 intent
Rule enginetable: engineer
Metric planner1 filter
Template compiler2 lines of SQL

query

SELECT summary, years, projects, certificates, users FROM engineer
  WHERE name = 'Abdullah Gökmen';

result 1 row

AI systems that turn plain-language questions into verifiable queries — inside public infrastructure serving 85 million people.

years of experience: 4+
projects shipped: 15+
certificates: 18
users served: 85M+

This box is a small in-browser model of MHRS IQ’s layered approach: no LLM, no server calls, every step visible.

record: person · 1 row

About

I’m an AI engineer at the Republic of Türkiye Ministry of Health in Ankara. I work on MHRS IQ, MHRS and ESİM: I design, build and scale systems that serve 85M+ users and need to stay up 24/7.

A system merely running isn’t enough; it needs to breathe under load, monitor itself, and heal itself when it has to. My own projects chase the same question: local memory for AI coding agents (Persistent Memory), a vault that never shows keys in plain text (Vaultic), and a monitoring and honeypot line for my own servers.

Working principles

Scalability Comes First
A system that collapses under load might as well never have existed. Architecture is traffic-ready from day one.
Observability
A system you can't monitor is a system you can't manage. Every service knows its metrics, every metric knows its alert.
Security Is the Default
Security isn't bolted on afterward; the line is built from day one with fail2ban, Keycloak, and strict session management.
AI Is a Force Multiplier
AI isn't a showcase feature, it's an engine part: verifiable, measurable, production-ready.

table: projects · 14 records

Projects

Selected work, from public-scale critical systems to my own products.

WHERE true → 14 rows

MHRS IQ Natural Language to Analytics

A layered AI platform that turns natural-language questions into verifiable analytics queries. Entity extraction, rule engine, metric planner, template compiler.
4-layer AIAI
- LLM
- Java 25
- Python
- ClickHouse
- PostgreSQL
- Qdrant
MHRS IQ is a layered AI analytics platform that lets you ask natural-language questions over healthcare data. The question first passes through an entity-extraction layer; a rule engine resolves context, a metric planner determines which measurements are needed, and a template compiler produces a verifiable analytics query. A BGE-M3-based Turkish embedding and a RAG running on Qdrant also power a semantic cache layer that speeds up similar queries. The result: no black box — every step is traceable, and answers are reproducible.
Input: natural language · Output: verifiable SQL
MHRS Central Physician Appointment System

The Republic of Türkiye Ministry of Health's official appointment system; high-traffic, 24/7 microservice architecture.
85M+ usersPublic sector
- Java 17
- Spring Boot
- PostgreSQL
- Redis
- Kubernetes
Serving more than 85 million citizens, this is the Republic of Türkiye Ministry of Health's official appointment system. A microservice architecture running 24/7 without interruption under heavy traffic; services scaled on Kubernetes, Redis-accelerated appointment flows, and a consistent PostgreSQL data layer. Development on the core team of one of Turkey's most critical public software systems.
24/7 uptime · 85M+ scale
Persistent Memory AI Memory Layer

Fully local persistent memory for AI coding agents: automatic decision/lesson extraction, hybrid semantic + keyword search, multi-agent AI Council.
Local & privateAI
- Python
- Ollama
- BGE-M3
- MCP
- SQLite
Persistent Memory is a fully local persistent memory layer for AI coding agents like Claude Code and Codex. It automatically extracts decisions and lessons from conversations, uses BGE-M3 embeddings on local Ollama, finds relevant context through hybrid semantic and keyword search, and feeds the agent's memory through fixed-budget automatic recall injection. It also includes a multi-agent AI Council discussion module and a web dashboard.
Vaultic AI Credentials Vault

A reference-based secrets vault that lets AI coding agents work with API keys without ever seeing them in plaintext.
Live platformAI
- TypeScript
- Infisical
- MCP
- Docker
- Node.js
Vaultic is a reference-based secrets vault that lets AI coding agents work with API keys without ever seeing them in plaintext. Built on an Infisical CE backend, agents work purely with vault references through a TypeScript MCP broker, an agent-vault proxy, and a CLI. Leak prevention is enforced through AI agent hooks (PreToolUse). Live in production at vault.abdullahgokmen.com.
Monitoring & Alerting Infrastructure

Prometheus + Grafana monitoring: system resources, containers, site uptime, and certificate lifespans on one dashboard; instant alerts on threshold breaches.
3 servers · 21 targetsSecurity & infra
- Prometheus
- Grafana
- cAdvisor
- Docker
A Prometheus + Grafana monitoring pipeline unifying 3 servers and 21 targets on a single dashboard. System resources, container metrics (cAdvisor), site uptime, and SSL certificate lifespans all on the same dashboard; instant alerts fire when defined thresholds are breached. Dashboards are accessible via grafana.abdullahgokmen.com and the infrastructure runs in production. No visibility means no system — this infrastructure leaves nothing in the dark.
Address Watch

A security pipeline that collects and classifies connection attempts to the server; honeypot + fail2ban integration.
Honeypot + fail2banSecurity & infra
- PostgreSQL
- Express
- Honeypot
- fail2ban
Address Watch is a security pipeline that collects and classifies every connection attempt to the server. Honeypot services lure attackers into a trap, attempts are logged to PostgreSQL and tagged by behavior pattern; fail2ban integration automatically blocks repeat threats. The attack surface is no longer invisible.
AINA AI Native Assistant

AI-assisted therapy session management for clinical psychologists; real-time voice transcription + AI analysis.
AI-poweredAI
- Next.js
- Node.js
- Whisper AI
- Claude API
- PostgreSQL
AINA is an AI-native assistant that manages clinical psychologists' session workflows end to end. During a session it captures real-time voice transcription (Whisper AI) and generates emotion and theme analysis via the Claude API; session notes, summaries, and follow-up plans are produced automatically. The clinician focuses on the conversation — AINA handles the paperwork.
BlackHoleLabs

Security-first, multi-tenant modular platform; a fully custom 13-module NestJS backend, automated domain/hosting bundle provisioning.
Live platformEnterprise
- NestJS
- TypeScript
- PostgreSQL
- Docker
- Stripe
BlackHoleLabs is a security-first, multi-tenant modular platform. Instead of off-the-shelf billing platforms, it's built on a fully custom NestJS backend split into 13 modules. The domain and hosting bundle provisioning process is automated, running live in production with 3 external integrations; mailcow also powers the platform's email infrastructure.
METUS Equipment Tracking System

Device tracking, maintenance scheduling, and data analytics platform.
Istanbul MetroPublic sector
- Microservices
- Docker
- Kubernetes
- Redis
Built for Istanbul Metro, METUS manages the lifecycle of distributed field devices: device tracking, periodic maintenance scheduling, and operational data analytics on a single platform. The microservice architecture runs on Docker and Kubernetes; Redis answers real-time field status queries in milliseconds.
Hotel Management System

ISO 22000 / HACCP-compliant, food-safety-focused hotel management application.
10+ hotelsEnterprise
- Spring Boot
- PostgreSQL
- React
- Docker
A management system actively used in more than 10 hotels, fully compliant with ISO 22000 and HACCP standards. Food-safety-focused inspection workflows, critical control point tracking, and reporting; a Spring Boot backend, a React operations panel, and standardized deployment via Docker. Inspection day means one-click reports, not panic.
FlexiPlay IPTV Media Player

Xtream API-powered multi-platform IPTV; mobile, web, and admin panel; subscription/payment management.
Multi-platformLive product
- Spring Boot
- Flutter
- React
- PostgreSQL
- Docker
FlexiPlay is a multi-platform IPTV solution powered by the Xtream API: a Flutter mobile app, a React web player, and a full-featured admin panel. Subscription and payment management runs on a Spring Boot backend; broadcast, user, and package operations all run from a single panel. Live in production and still gaining subscribers.
Boostly SMM Panel

298+ service packages, 11+ social platforms; digital marketing automation panel.
LiveLive product
- FastAPI
- Next.js
- PostgreSQL
- Stripe
- Docker
Boostly is an SMM system bringing together more than 298 service packages and more than 11 social media platforms in a single automation panel. A high-speed order engine built with FastAPI, a smooth Next.js panel, and secure payments via Stripe. Digital marketing operations are automated end to end; live in production and scaling.
Mobile App Projects

Cross-platform mobile applications built in Flutter and shipped to market.
5K+ downloadsLive product
- Flutter
- Dart
- Firebase
- REST API
A series of cross-platform mobile applications built with Flutter and successfully sold. A single codebase producing both iOS and Android builds, with Firebase for real-time data and REST API integrations. More than 5,000 total downloads — proof of the path from idea to app store.
AIGram AI Social Network

A social network where AI agents interact autonomously; posts, stories, DMs, webhook infrastructure.
ArchivedAI
- Go
- Next.js
- PostgreSQL
- Redis
- Docker
AIGram was built as a social network where AI agents interact with each other autonomously. Agents post, share stories, and send DMs; external systems connect to the feed through a webhook infrastructure. A high-performance Go backend, a real-time Next.js interface, and a Redis-backed feed engine delivered an end-to-end autonomous social experience.

table: experience · 4 records · 2021 — today

Experience

Republic of Türkiye Ministry of Health
HBTech
Abdullah Gökmen Sole Proprietorship
Olbia Bilişim Teknolojileri

October 2024 — today 1 yr 11 mos

Republic of Türkiye Ministry of Health

AI Engineer

I run the software support processes for the MHRS, ESİM, and Software Support applications. I'm responsible for scalability, versioning, and 24/7 uninterrupted availability in high-traffic public applications.

Systems MHRS IQ · MHRS · ESİM · Software Support
- Java 17
- Spring Boot
- PostgreSQL
- Redis
- RabbitMQ
- React
- Docker
- Kubernetes
- Jenkins
- Grafana
August 2023 — August 2024 1 yr

HBTech

Software Developer (Full-Time)

I developed ISO 22000 and HACCP compliant hotel management systems. I worked on the METUS Equipment Tracking System for the Istanbul Metro, and delivered end-to-end development on custom software projects.
- Flutter
- Spring Boot
- PostgreSQL
- React
- Docker
- Kubernetes
- Microservices
- GitLab CI/CD
January 2022 — August 2023 1 yr 7 mos

Abdullah Gökmen Sole Proprietorship

Freelance Software Developer & Entrepreneur

I built e-commerce platforms, handled Linux system administration, and developed and sold custom software solutions for SMEs' digital transformation.
- E-commerce
- Linux
- System administration
- PostgreSQL
- MySQL
September 2021 — January 2022 4 mos

Olbia Bilişim Teknolojileri

Software Developer (Long-Term Intern)

I developed modern frontend interfaces and RESTful APIs, and worked on secure session management and agile development processes.
- React.js
- Vite
- Java Spring Boot
- Node.js
- PostgreSQL
- MongoDB
- Docker
- JWT
- Swagger

table: skills · 16 records · 5 groups

Skills

Levels: Expert, Advanced, Operational. Run any row in the query box above.

Backend

Java Used in: MHRS · MHRS IQ · FlexiPlay
Spring Used in: MHRS · METUS · Hotel
LLM / AI Engineering Used in: MHRS IQ · AINA · Vaultic
Python Used in: MHRS IQ · Persistent Memory
JavaScript Used in: React · toolchain
gRPC Used in: Inter-service communication

DevOps & infrastructure

Docker Used in: All deployments
Linux / Ubuntu Used in: All servers and deployments
Jenkins Used in: METUS · deploy pipelines
Git Used in: Every repo, every day

Data

PostgreSQL Used in: MHRS · Hotel · FlexiPlay
Redis Used in: MHRS · METUS
Elasticsearch Used in: MHRS · Address Watch

Frontend

React Used in: MHRS IQ · Boostly

Security & messaging

Kafka Used in: MHRS · AIGram
Keycloak Used in: MHRS · METUS

table: certificates · 18 records · by date

Certificates

Certificate	Issuer	Date	Credential ID
Introduction to Programming with JAVA	BTK Akademi	November 2022	`2NwcOYqJkN`
DoS / DDoS Attacks and Protection	BTK Akademi	August 2022	`PvghbYlXq1`
Google Project Management	Coursera	July 2022	`7J4VYGH8PJQ6`
Game and App Academy (Graduation)	Google	July 2022	Graduation Certificate
Technology Entrepreneurship	Google	July 2022	`66371185908780`
Game Art Training	Google	July 2022	`1549758000385`
App Development	Google	July 2022	`64842503391725`
Agile Project Management	Coursera	June 2022	`ZUWULXE5UQJ9`
Capstone: Applying Project Management in the Real World	Coursera	June 2022	`HPXH9WXGCTEW`
Project Execution: Running the Project	Coursera	April 2022	`5VDCG3MVUUDG`
Project Planning: Putting It All Together	Coursera	March 2022	`EZFXCDBB3APW`
Google DSC — Flutter Week	Google DSC	February 2022	`04381057328061`
Mobile App Development with Google Flutter	BTK Akademi	January 2022	`vpWc8zwoDY`
Project Initiation: Starting a Successful Project	Coursera	January 2022	`MBLWDAY9DQ6`
Foundations of Project Management	Coursera	January 2022	`263EL2GLVS9F`
Introduction to Digital Game Development with Unity	BTK Akademi	June 2021	`d6GQZ63L`
Coding for Web Design	Anadolu Üniversitesi	October 2020	`MY0UCONIIV`
Applied Entrepreneurship Training	KOSGEB	November 2018	Certificate of Participation

endpoint: POST /api/contact

Contact

Open to freelance projects, consulting and full-time opportunities. I usually reply within 24 hours.

Direct

Request that will be sent

POST /api/contact
Content-Type: application/json

{
  "isim": "",
  "eposta": "",
  "konu": "",
  "mesaj": ""
}

What you type is sent only in this request and delivered as an email.

Abdullah Gökmen — AI engineer

About

Working principles

Scalability Comes First

Observability

Security Is the Default

AI Is a Force Multiplier

Projects

MHRS IQ Natural Language to Analytics

MHRS Central Physician Appointment System

Persistent Memory AI Memory Layer

Vaultic AI Credentials Vault

Monitoring & Alerting Infrastructure

Address Watch

AINA AI Native Assistant

BlackHoleLabs

METUS Equipment Tracking System

Hotel Management System

FlexiPlay IPTV Media Player

Boostly SMM Panel

Mobile App Projects

AIGram AI Social Network

Experience

Republic of Türkiye Ministry of Health

HBTech

Abdullah Gökmen Sole Proprietorship

Olbia Bilişim Teknolojileri