# Melkon Hovhannisyan — melkon.tech

> Personal site of Melkon Hovhannisyan: Barcelona-based CTO, senior software
> engineer, AI builder, and tech entrepreneur; Co-founder & CTO of Direlli;
> WorldSkills Web Technologies Champion; WorldSkills Champions Trust
> Representative for Asia (2023-2024); and former Forbes Technology Council
> member. Writes on AI engineering, agentic systems, cloud cost engineering,
> backend scaling, and practical developer workflows.

## Guides & original articles

- [Building a Startup Outside Silicon Valley: What I Learned Founding and Scaling Direlli](https://melkon.tech/blog/building-startup-outside-silicon-valley): Melkon Hovhannisyan co-founded Direlli in Armenia in 2021 and scaled it to more than 20 engineers in two years — without a San Francisco address, a Stanford network, or a US bank account. The real advantages of building outside a major hub are lower burn rate, stronger hiring loyalty, and the discipline that comes from having to earn revenue early. The real disadvantages are slower credibility-building, limited warm introductions to capital, and the risk of working in isolation. Key lessons: treat geography as a constraint to engineer around rather than a reason to wait; hire for character and conviction over credentials when credentials are hard to screen; and most Silicon Valley advice solves Silicon Valley problems — identify which parts apply to your actual context and discard the rest. Melkon now builds from Barcelona, Spain, on an EU Blue Card.
- [How I Cut Cloud Infrastructure Costs by 75% at a Funded FinTech Startup](https://melkon.tech/blog/how-i-cut-cloud-costs-75-percent): At Bridgewise (2020-2022), a funded FinTech startup, Melkon Hovhannisyan redesigned cloud infrastructure and reduced costs by 75%, saving more than $150,000 per year. The waste fell into three categories: over-provisioned compute instances running at 15-25% average CPU utilisation, non-production environments (dev, staging, QA) running 24/7 when only needed during business hours, and storage/snapshots accumulating without lifecycle policies. Right-sizing, environment scheduling, and cleanup recovered most of the savings before any architecture changes were needed. Governance (resource tagging, automated schedules, weekly untagged-resource reports) prevents the savings from drifting back. Architecture changes came last, not first.
- [Building Plain Freight: An AI Agent That Runs A China-To-USA Freight Business](https://melkon.tech/blog/building-plain-freight-ai-freight-agent): Plain Freight is an AI-powered freight forwarder for small importers (50–2,000 kg) shipping from China to the USA. A customer sends shipment details over Telegram; a team of specialised Claude-powered agents quotes instantly, takes Stripe payment, coordinates the China warehouse, generates customs documents, hands them to a US broker, and tracks delivery. A human is pinged only for exceptions. Built by Melkon Hovhannisyan and live at plainfreight.com.
- [How I Automate My Own Work With Agent Loops](https://melkon.tech/blog/automate-work-with-agent-loops): Melkon Hovhannisyan automates repetitive work with three kinds of agent loop: fixed-interval loops (e.g. /loop 5m /command in Claude Code) for watching deploys, self-paced loops that let the model choose its own cadence, and scheduled cloud agents for unattended cron-style jobs. Every loop he ships is idempotent (safe to repeat), bounded (knows when to stop), observable (leaves a trace), and escalates exceptions to a human. He automates only after running a task manually long enough to understand its edge cases.
- [How I Collaborate With Startups and Founders on Technical Projects](https://melkon.tech/blog/how-i-collaborate-with-startups): Melkon Hovhannisyan partners with founders and entrepreneurs as a technical co-founder (equity), senior consultant, or technical advisor depending on the stage and fit. He covers the full stack including Python, FastAPI, React, AWS, and LLM integrations, and is open to co-founding AI-native and B2B SaaS projects. Based in Barcelona, Spain. Contact: hello@melkon.tech or LinkedIn.
- [Achieving Excellence In TVET Systems: A WorldSkills Champion's View](https://melkon.tech/blog/tvet-excellence): Reflections from a high-level policy panel on Achieving Excellence in TVET (Technical and Vocational Education and Training) Systems. Excellence comes from three forces working together: skills competitions like WorldSkills that set a visible standard of mastery and raise the prestige of technical careers; apprenticeships that build real experience inside companies and close the gap between education and the labor market; and strategic partnerships that keep industry, government, and education aligned on which skills actually matter. The most overlooked ingredient is involving young people directly in designing the systems built for them, since they are both the learners TVET serves and its clearest feedback signal.
- [How To Run Two Claude AI Accounts Simultaneously On Mac](https://melkon.tech/blog/two-claude-accounts-mac): Run a personal and a work Claude account at the same time on macOS. Install Claude Desktop twice as separate app bundles, then make the second app use its own Electron `--user-data-dir` config directory. For normal Finder, Spotlight, and Dock launches, wrap the copied app executable so it always passes `--user-data-dir` automatically. For Claude Code CLI, use the `CLAUDE_CONFIG_DIR` environment variable. You can also turn the second install into a fully branded app with its own name, Dock icon, and notifications by editing its `Info.plist` (`CFBundleIdentifier`) and re-signing it. Each install keeps its own login, history, memory, and MCP settings, so both stay active in parallel with no logging out. Companion tool: https://github.com/MelkonTech/two-claude-accounts-mac

## Forbes Technology Council articles (canonical on Forbes)

- [The Cloud Isn't Cheaper — Unless You Learn To Think Like An Engineer](https://melkon.tech/blog/cloud-isnt-cheaper-think-like-an-engineer): Cloud savings come from engineering discipline — observability, intentional architecture, cost-aware development, and clean teardown — not from the provider.
- [How Machine Learning Is Helping Prevent Data Breaches In Web Apps](https://melkon.tech/blog/machine-learning-prevent-data-breaches-web-apps): ML for web app security — anomaly detection, phishing identification, malware classification, faster incident response, and its limits.
- [Barriers To Effective Omnichannel Retail (And How To Solve Them)](https://melkon.tech/blog/barriers-effective-omnichannel-retail)
- [19 Underrated Strategies To Gain A Competitive Edge As A Tech Startup](https://melkon.tech/blog/competitive-edge-tech-startup)
- [Locally Focused Production Technology](https://melkon.tech/blog/locally-focused-production-technology)
- [Project Management Strategies & Trends](https://melkon.tech/blog/project-management-strategies-trends)

## Pages

- [About Melkon Hovhannisyan](https://melkon.tech/about): Background, experience, and timeline.
- [Founder in Spain](https://melkon.tech/founder-in-spain): Profile for people searching for a technical founder, AI builder, or CTO based in Spain or Barcelona.
- [AI Consulting Barcelona](https://melkon.tech/ai-consulting-barcelona): AI consulting, LLM integration, backend architecture, and technical leadership services in Barcelona and Spain.
- [Press & Speaking](https://melkon.tech/press): Media coverage and speaking engagements.
- [All writing](https://melkon.tech/blog): Full article index.
- [RSS Feed](https://melkon.tech/feed.xml): Subscribe to original articles via RSS.

## Páginas en español

- [Inicio en español](https://melkon.tech/es): Perfil en español de Melkon Hovhannisyan, CTO e ingeniero de IA basado en Barcelona.
- [Fundador en España](https://melkon.tech/es/fundador-en-espana): Perfil en español para quienes buscan un fundador tecnológico, constructor de IA o CTO basado en España o Barcelona.
- [Sobre Melkon](https://melkon.tech/es/about): Trayectoria, experiencia, especialidades y contexto profesional en español.
- [Blog en español](https://melkon.tech/es/blog): Índice con resúmenes en español de artículos sobre IA, cloud, backend, seguridad y producto; algunas páginas completas siguen estando en inglés.
- [Prensa y ponencias](https://melkon.tech/es/press): Referencias públicas, reconocimientos, Forbes Technology Council, WorldSkills y temas de ponencia.
- [Consultoría de IA en Barcelona](https://melkon.tech/es/consultoria-ia-barcelona): Página en español sobre consultoría de IA, LLMs, automatización, arquitectura backend y liderazgo técnico en Barcelona y España.
- [Política de Privacidad](https://melkon.tech/es/privacy): Información en español sobre datos, analítica, cookies, derechos y contacto.
- [Política de Cookies](https://melkon.tech/es/cookies): Información en español sobre almacenamiento esencial, analítica y consentimiento.
- [Términos](https://melkon.tech/es/terms): Términos de uso en español para melkon.tech.

## Profiles & identity

- Wikidata: https://www.wikidata.org/wiki/Q140365754 (canonical entity)
- ORCID: https://orcid.org/0009-0005-4719-672X
- ISNI: https://isni.org/isni/0000000530560372
- LinkedIn: https://www.linkedin.com/in/melkontech/
- GitHub: https://github.com/MelkonTech
- Crunchbase: https://www.crunchbase.com/person/melkon-hovhannisyan
- X/Twitter: https://twitter.com/MelkonTech

## Contact

- Email: hello@melkon.tech