AI Prompt Evaluator — TELUS Digital
05/2025 - 03/2026
- Evaluated LLM outputs for accuracy, safety, and instruction-following.
- Refined prompts and interaction flows to improve reliability and reduce ambiguity.
Available for new projects · Europe-based
Production AI systems
Automated precision.
Enterprise AI Systems Engineer (RAG, Agents, Automation) I build production AI systems with grounded retrieval, deterministic workflows, guardrails, and secure delivery for serious B2B teams.
System snapshot
OperationalPipeline reliabilityValidation-first
Compliance posturePrivacy-aware
Delivery modelB2B production
RAG • Agents • Automation in real constraints
Experience
Independent AI Systems Engineer — Self-employed
2024 - Present
- Built production-style AI systems across RAG, document analysis, compliance workflows, and backend automation.
- Focused on deterministic pipelines, privacy-aware processing, and reliable AI-assisted outputs.
Selected work
Applied AI systems engineered for production: grounded retrieval, validation-first orchestration, security boundaries, and maintainable delivery for B2B and enterprise teams.
Live product
FileGPT.dev
Architecture
Tenant-isolated RAG orchestration
Reliability
Grounded retrieval, citations, and access control
Business
Trusted internal knowledge answers
Problem: Internal knowledge was spread across files and inboxes, so teams lost hours searching for answers and still doubted source reliability.
Solution: FileGPT.dev is a private document Q&A vault with account-scoped access, grounded retrieval, citation-backed answers, and cost-aware orchestration for predictable operations.
Impact: Teams get faster answers without sacrificing trust, because outputs stay reviewable, source-backed, and aligned with internal access boundaries.
Next.js 15SupabaseRAGVercel AI SDKGeminipgvector
Live product
TrustRespond.ai
Architecture
Document-to-Excel RAG orchestration
Reliability
Validation-first output + formatting integrity
Business
Faster, reviewable security responses
Problem: Enterprise B2B deals stall on massive vendor security questionnaires—often 200-row Excel files—while teams spend weeks mapping SOC 2 reports and internal policies into spreadsheets that break and lose context.
Solution: TrustRespond.ai ingests compliance documents into pgvector, maps questions through an enterprise RAG pipeline, and returns reviewable Excel outputs while preserving workbook structure.
Impact: Typical runs finish in about 12 seconds instead of weeks, helping teams respond faster while preserving reviewability and B2B trust.
Next.js 15SupabasepgvectorGemini 2.5StripeTailwind CSS
Featured project
Live product
ComplianceRadar
Architecture
Guarded URL ingestion + compliance scoring pipeline
Reliability
Policy-grounded reports with abuse-aware controls
Business
Faster AI Act/GDPR decision cycles
Problem: SMBs face rising AI Act, GDPR, and ePrivacy risk but cannot justify expensive enterprise compliance programs.
Solution: ComplianceRadar combines guarded URL scanning, policy-grounded AI analysis, and actionable reporting so teams can triage EU AI Act, GDPR, and ePrivacy risk without legal ops overhead.
Impact: Product and leadership teams move from uncertainty to prioritized remediation faster, while keeping launches aligned with compliance expectations.
Next.jsGemini AIPrismaStripeNextAuthVercel
Additional Projects
Additional systems and case studies available on request.
Hausheld
Impact: Teams reduced coordination overhead, improved service reliability, and gained audit-ready documentation for regulated care delivery.
Next.jsFastAPIPostgreSQLAWSPostGIS
Croatia 360
Impact: Users plan faster with less friction, discover more relevant options, and convert inspiration into completed travel decisions.
Next.jsi18nGoogle GeminiTailwind CSS
How I build systems
AI workflow architecture
A lightweight view of how I structure production-ready AI systems: deterministic flow control around LLM intelligence, with validation and observability built in.
User input
Requests enter through typed interfaces with schema-safe parsing and context capture to avoid ambiguity at the edge.
Reliability cues
Traceable steps and outputs
Validation before delivery
Policy-aware orchestration
Interactive Simulation
Interactive RAG Console
Test-drive the core RAG extraction logic that powers my systems. Select a target B2B application and query below to see the animated execution pipeline, guardrails, and citation-grounded output.
Orchestrator Terminal
rag-orchestrator.service: inspector
ONLINE
Select an AI system and predefined query, then click 'Execute RAG Pipeline' to run the interactive RAG visualizer and see active execution logs in real-time.
Want a system like this for your team?
I specialize in AI-powered systems, GDPR-ready platforms, and production-grade full-stack applications for European businesses. Tell me what you're building and I'll respond within 24 hours.
Skills
Backend
PythonNode.jsFastAPISQLAlchemyPostGISAlembicPrismaPostgreSQLPydanticREST APIsJWT
AI & Data
RAGOpenAI APIGoogle GeminiPineconeVercel AI SDKEmbeddingsVector searchConversation memory
Frontend
Next.jsReactTypeScriptTailwind CSSPWAi18nFramer MotionShadcn/RadixRecharts
DevOps
DockerAWS (eu-central-1)Google CloudVercelCI/CDGit
The Blueprint
Inspect the portfolio codebase directly in this mock VS Code environment. Click the files in the workspace explorer to read how different architectural components are engineered.
damir@andrijanic-dev: ~/portfolio-2026
index.ts
// Damir Andrijanic — Portfolio entrypoint
import { RagOrchestrator } from "./rag-orchestrator";
import { XlsxMapper } from "./xlsx-mapper";
export const engineer = {
name: "Damir Andrijanic",
title: "Enterprise AI Systems Engineer",
focus: ["RAG", "Agent Orchestration", "Deterministic Automation"],
regions: ["Germany", "Croatia", "European B2B"]
};
export async function runSimulation() {
console.log("Automated precision.");
const orchestrator = new RagOrchestrator();
await orchestrator.execute("Audit GDPR compliance posture");
}
About
I build enterprise-ready AI systems focused on RAG, agent orchestration, and automation under real production constraints.
My approach is validation-first: deterministic rules and system boundaries where needed, with LLM reasoning layered on top for speed and explainability. The objective is reliable outcomes, not demo output.
I work on systems where retrieval quality, citations, abuse controls, rate limits, privacy posture, and maintainability matter as much as model quality.
Most of my work sits at the intersection of backend architecture, data pipelines, and applied AI product engineering for B2B and enterprise use cases.
I came to tech through a non-traditional path, which shaped how I engineer systems: understand the real workflow first, then ship software that holds up in production.
Get in touch
Hiring for an AI Systems Engineer or Applied AI role, or need help shipping a production AI workflow? I work with recruiters, hiring managers, founders, and technical leads, and usually reply within 24 hours.