How I work. Named outcomes, no theater.
Mostly hired in-house as a full-stack web developer. Occasionally available through Toptal to build AI products: SaaS MVPs, LLM pipelines, internal tooling, and the integrations that hold them together.
AI tooling that survives the prompt-to-production gap.
Most AI projects die on the way from notebook to running system. I build the in-between.
LLM Pipelines & Agent Prototypes
The hiring artifact, built as a service: ingestion, classify, summarize, ship.
- Production agent pipelines, local-first or cloud, with durable storage and observable cost
- Model routing via OpenRouter; cost and latency tracked via Helicone
- Storage on Turso, Postgres, or pgvector; cron on Railway or your existing infra
- Prompt management with versioning, evals, and team-shareable libraries
- The same shape I ship in CIA, production-grade, not a demo notebook
Answer Engine Optimization
Beyond SEO. Get cited by ChatGPT, Perplexity, and Gemini, not just ranked by Google.
- JSON-LD schema (Person, FAQPage, Article, CollectionPage) across templates
- Question-focused headings and answer-paragraph patterns that AI parses cleanly
- Entity and authority establishment, the bits AI uses to decide whether to cite you
- AEO audits with concrete next steps, ranked by effort and impact
- Implemented directly in HubSpot CMS, WordPress, or your custom Next.js stack
Web & Integration Engineering
The full-stack web and integration work production AI still runs on.
- Full-stack web on Next.js and React, plus WordPress and HubSpot CMS when right
- Internal dashboards and tools teams own without an engineering ticket
- Integrations: REST APIs, webhooks, n8n, and your custom systems
- Performance, Core Web Vitals, accessibility (WCAG), and AEO baked in
- Production code review and documentation, no consultant theater
Embedded Engineering with AI
Drop in for a sprint or a quarter. Ship like a team member, AI velocity baked in.
- Joins your standups, code review, and Slack, no account-manager middle layer
- Production code review, security, and accessibility defaults from day one
- Documentation that survives team turnover, and is actually written down
- AI-assisted velocity with hardening: prototype fast, ship production-grade
- Best when the role is new, the title is fuzzy, or the team needs an AI-native voice
How every engagement runs.
Direct, no middle layer
You work with me, not an account manager. Decisions in hours, not next week's call.
Enterprise defaults
Strict code review, security-first, WCAG, semantic HTML, type-safety, the same standards Cognigy and Regiondo expected.
Documentation that ships
Handover docs your team will actually open, built so I am not the single point of failure.
Questions clients and recruiters actually ask.
What does Aleksandar Perisic build?
Full-stack web and AI products: multi-tenant SaaS, LLM pipelines and agent systems, vector search and auth, realtime features, plus the internal tooling and integrations around them, with fifteen-plus years of full-stack web underneath. He works as a full-stack web developer at NiCE Cognigy and takes select engagements through Toptal.
What services does Aleksandar Perisic offer?
Building AI products (SaaS MVPs, LLM pipelines, agent systems), Answer Engine Optimization (AEO), internal tooling and dashboards, web and CMS development, n8n workflow automation, and technical architecture reviews. Stack includes Claude Code, OpenRouter, Helicone, Turso, Postgres/pgvector, Qdrant, Railway, n8n, React and Next.js.
What is Answer Engine Optimization (AEO)?
Answer Engine Optimization is the practice of structuring website content so it appears in answers generated by ChatGPT, Perplexity, Gemini, and other AI assistants. It builds on classic SEO but prioritizes semantic clarity, structured data (JSON-LD), question-focused headings, and authoritative citations that AI systems can parse and reference directly.
Where is Aleksandar Perisic based?
Near Munich, Germany. Currently full-time at NiCE Cognigy and available for select freelance engagements via Toptal, and open to remote senior in-house roles with B2B companies building AI products.
What is Aleksandar Perisic's daily AI stack?
Claude Code as the primary coding assistant, OpenRouter for model routing, Helicone for cost and latency tracking, n8n for workflow automation, Turso for edge storage, Coolify and Railway for self-hosted infrastructure. Models are switched per task, no model loyalty.
Need an AI build that holds up?
Based near Munich, open to senior in-house roles, remote in Germany, plus select Toptal engagements.
Prefer email? hi@aleksandarperisic.com.