Software Development Products: sometoo.de, coinanalyst.tech, cogia.de, memonews.de, influverse.io, elastic.io Servers(OS): 200+ Debian, Ubuntu, CentOS, Windows Server 2012, 2019 Hosting providers: Hetzner Cloud/Robot , GCP, AWS Services: MariaDB, MySQL, nginx, Apache HTTP server, monitoring, InfluxDB, Apache Tomcat, Apache Solr, Gitlab, Cassandra, Zookeper, Kubernetes, Deckhouse(k8s platform from Flant), Kafka, Docker, MS Exchange, Artifactory, Pritunl Programming: Python, Flask, Java(small fixes), Various REST APIs (e.g. Hetzner, Cloudflare, Gitlab), Golang, JS Software: Ansible, Docker, Git, Prometheus, Grafana, Alerta, Zabbix, Java 11 + Maven, OpenVPN, Terraform, Helm Tasks: Automation, Troubleshooting, Services reliability, Development(small services), CI/CD pipelines Goals: Main goal is to make everything automated, idempotent, observable and documented. Subgoals: 1. Not even improve. Create observability. There were no metrics and logs gathering 2. Automate ops tasks like user management and server preparation with Ansible 3. Document everything 4. Describe every service deploy with Ansible/k8s yamls. Achievements 1. Monitoring by-default. Every new server and service automatically has prometheus exporter registered at consul. Various dashboards for quick investigations present. Every past occured issue has alert and reacted. 2. All databases and uploads are backed up with restic. Ansible role for mysql backups is embedded to mysql server installation role 3. CI/CD for 95% of services changed from often-failing bash scripts to idempotent ansible playbook runs. Creating one new instance of existing app takes 1-4 hours instead of 1-4 days. Deployment to production does not even need my presence. Previously it was a lottery: would it deploy and run successfully on production or not. 4. Every repository which has app running somewhere on server does have `install` directory with all required inventories and playbooks. Changes on servers are done via pushing changes to roles/playbooks in repository. Previously there were 100+ manually-configured `snowflake` servers.

Enterprise Network: 10+ Cisco, MikroTik, 3Com, SNR, D-Link Servers(OS): XenServer, Proxmox VE, Debian, Ubuntu, CentOS, Oracle Linux, FreeBSD, Windows Server 2012-2016 Servers(HW): Dell, HP, Fujitsu, Intel, Huawei Dorado Services: MariaDB, MySQL, nginx, apache http server, rocketchat, asterisk, monitoring, Active Directory Programming: Python, Django, vuejs, PHP, jQuery, Electron, PowerShell, scikit Software: Ansible, Docker, Git, Prometheus, Grafana, Alerta, FreePBX, Zabbix, iSCSI(multipath) Tasks: Automation, Integration, Troubleshooting, Internet load balancing and failover, Services reliability, AI(Machine learning, Speech Recognition, Regression analysis, Classification) Goals: 1. Improve stability. (Proxmox+SAN+HA, Electrical load distribution, Aggregation and failover of internet channels) 2. Increase obsrvability. (Prometheus w/autodiscovery+Grafana+Dashboards) 3. Automate routine operations (50+ self-written small simple tools with self-written GUI for them) Achievements: 1. Downtime of important services decreased from hours to minutes 2. Time to find root cause of complicated problems derceased from 2-3 hours to 15-30 minutes. Also now we are able to see overall hardware load quickly and choose correct place for new services to balance load between HW 3. Many routine operations like "grant access to new employee" now performed with one button and 1-2 minutes of time. Every operation took 15-30 minutes earlier

ISP Network: 100+ Ubiquiti, 20+ MikroTik, D-Link, SNR Servers(OS): Proxmox VE, Debian, FreeBSD, CentOS, Windows Server 2008 Services: Billing, monitoring, asterisk, MySQL, apache http server, online payments Programming: PHP, Bash, Python Software: Ubilling, Zabbix, FreePBX Tasks: Troubleshooting, Automation, Integration, Support When I started to work - accounting was in Excel table, company's clients were performing "alerting" function of monitoring via mobile phone. Troubleshooting was taking MUCH time. And if someone did not pay for several month - CTO was logging in to BRAS and manually disabling this client's connection Goals: 1. Create billing system. Automate everything related to accounting and billing 2. Create monitoring and alerting system so we are able to know about problems before clients "alert" us =) 3. Make something with phone. Achievments: All goals were achieved. No more manual disconnections, all systems became observable and company's phone became PBX with redirections, queues and IVR And one more thing: when you start to automate - it's really hard to stop. So when client called me and asked what is wrong with his Internet connection, I was able to see who is calling(PBX+Billing integration) and know what exactly is wrong(billing+network hardware integration)

Tech Support Network: Juniper(EX, SRX series) Servers(OS): CentOS, FreeBSD(pfSense) Services: asterisk, terminal server