I'm Daniel Semerjyan, a Big Four technology consultant and hands-on AI engineer. I help enterprises adopt AI-native ways of building software, and I practice what I advise: shipping RAG, agentic, and GenAI products end to end on AWS and Azure.
Everything I do is one craft, AI-native development, applied at different altitudes: the enterprise, the product, the team.
For enterprises
Strategy that survives contact with delivery. I help organizations move from AI pilots to AI-native development: the tooling, workflows, and engineering culture that make it stick.
For founders & product teams
RAG pipelines, agentic systems, GenAI features. Designed, built, and deployed on AWS or Azure. Production-grade from day one: auth, observability, and infrastructure included.
For leaders & teams
I train teams to build with AI instead of around it, review architectures before they calcify, and give the kind of direct feedback consulting decks usually soften.
Products I designed, built, and shipped. Every one is AI-native in how it was made, several in what they do.
AI proposal-generation SaaS for service businesses — turns a client inquiry into a polished, on-brand offer in minutes.
Compliance automation for Polish real estate developers — auto-generates daily price reports to the dane.gov.pl portal from a single listing URL.
Marketing site for an insurance and financial advisory practice serving the Polish transport sector.
Fitness management platform for personal trainers — schedule clients, build recurring sessions, and manage workouts from a single dashboard.
Chrome extension that records any browser tab (+ mic) and produces speaker-labeled markdown transcripts — locally, with no subscription or remote server.
Essays on AI-native development, agents, and building products. Applied frameworks and case studies from the work.
Module libraries rot because standards live in prose and reviewers enforce them from memory. This repo turns every standard into a machine-checkable gate, then lets a coding agent do the work. A tour of the harness, and what it changes in a platform team's week.
The bottleneck in AI work is the human in the middle handing out tasks. Long-running autonomous agents are the specific mechanism that removes you from the loop, and as of Opus 4.8 they ship rather than loom. The optimist's case for climbing to Level 3.
Managed Postgres at small scale starts at $25–70 per project per month. There's a middle ground the debate usually skips: one cheap VPS, multiple projects, real backups, real DR, ~$7/mo total. This is the recipe.
AI agents now decide who gets seen on the web. The peer-reviewed research is in, and the playbook is clearer than most posts make it sound — six moves, in priority order, for anyone running a brand, a portfolio, or a product.
I work with enterprises, founders, and product teams to take AI-native ideas from ambiguous to shipped. The consulting side brings scoping, business-model assessment, and structured delivery; the engineering side brings a decade of web platforms, backend systems, and cloud.
Honest feedback runs across all four steps, not just the first one.
A working session to pin down the use case, sanity-check the business model around it, and find where AI adds leverage and where it doesn't.
An MVP in days, built with AI for speed, so we can put it in front of a real user and learn before committing to a roadmap.
Deployed, observable, in real users' hands. A product with auth, infra, and the operational basics in place.
Harden, optimize, and extend. Default stack: Next.js, with AWS or Azure underneath and the right AI primitives for the job.
Not collected for the wall. Earned building the systems the certificates describe.
Amazon Web Services
4× certifiedMicrosoft Azure
5× certifiedWhether you're an enterprise leader planning AI adoption, a founder who needs a product shipped, or a team that wants to move faster, the quickest way to find out if I can help is a short email.
Poznań, Poland · Working across EU time zones
