Andrej Ilievski
Verified Expert in Engineering
Data Scientist and AI Developer
Bitola, Municipality of Bitola, Macedonia
Toptal member since August 25, 2025
Andrej is a senior AI engineer with 6+ years building production-grade AI systems for Fortune 500 clients in healthcare, real estate, and automotive. He specializes in Agentic AI, shipping AI agents for live HIPAA-compliant healthcare call analysis on Azure and automating real estate marketing generation AI tools, alongside GraphRAG pipelines and GNN-based predictive maintenance. He's proficient across Azure, AWS, and GCP, and backed by four IEEE, ACM, and Springer peer-reviewed publications.
Portfolio
Experience
- Python - 6 years
- Machine Learning - 5 years
- Amazon Web Services (AWS) - 5 years
- PyTorch - 2 years
- AI Agents - 2 years
- Graph Neural Networks (GNNs) - 2 years
- GraphRAG - 1 year
- Google Cloud Platform (GCP) - 1 year
Preferred Environment
Python, PyTorch, PyTorch Geometric (PyG), Amazon Web Services (AWS), Google Cloud Platform (GCP)
The most amazing...
...: I've shipped production AI agents in healthcare and real estate, pioneered GNN predictive maintenance in automotive, and secured $5M in Fortune 500 funding.
Work Experience
Senior AI Engineer
Bottimmo AG
- Designed and shipped a generator-discriminator agentic AI system for automated marketing content generation, enforcing content diversity via max cosine similarity (<0.65) validated across 50+ controlled experiments.
- Reduced CI/CD pipeline build time by 99% (from around 2 minutes to around 20ms) by migrating to uv and replacing legacy Ruby/dpl, eliminating redundant test executions across branch, main, and deployment stages.
- Engineered a REST API for personalized AI photo generation using Gemini nanobanana, enabling users to submit profile photos and receive realistic AI-generated portraits at scale.
AI Consultant
Astoria Health PLLC
- Designed and developed an AI agent for live transcription processing and generating AI-driven insights of healthcare calls.
- Architected real-time speech-to-text infrastructure using Azure Speech Services and WebSocket protocol for a US healthcare client, enabling HIPAA-compliant transcription optimized for sub-second latency.
- Designed an Azure-native event pipeline coordinating Web PubSub, Service Bus, and App Service to broadcast live healthcare transcripts to the front end — spanning compute, messaging, and real-time communication layers.
- Orchestrated a multi-service system spanning RingCentral webhooks, Python/FastAPI audio server, and TypeScript AI back end — decoupled via Azure Service Bus for fault-tolerant, scalable processing across transcription and insights stages.
Senior Data Scientist
Gymshop.mk
- Architected, developed, and deployed a consumer-facing AI chatbot leveraging GraphRAG architecture, utilizing LangChain, Chroma DB, Vertex AI, and GraphDB with NebulaGraph.
- Engineered an efficient product data flow using Python ETL processes, ensuring consistent formatting and seamless integration of disparate data sources into WordPress.
- Implemented custom event tracking in Google Analytics, generating actionable insights that improved product strategy execution.
- Integrated a fully serverless, always-on solution on Google Cloud, incorporating robust security measures across the web-facing application and back-end services.
- Optimized system performance, consistently achieving response times under five seconds.
Data Scientist
Symphony
- Spearheaded an industry-first predictive maintenance deep learning system using graph neural networks (GNNs), taking it from proof of concept (POC) to production and enhancing maintenance capabilities for a major automotive manufacturer.
- Achieved a 20% improvement in prediction accuracy over baseline neural networks through innovative GNN implementation, demonstrating superior performance for critical automotive applications.
- Developed a custom synthetic data generation methodology, improving GNN prediction accuracy by 10% and increasing the predictability of previously undetectable failure classes by 15%.
- Conducted comprehensive exploratory data analysis, directly securing project funding from a Fortune 500 automotive manufacturer by demonstrating the business value and ROI potential of GNN models.
- Optimized business intelligence dashboards using Python, Pandas, and Tableau, boosting performance by 40% and enhancing stakeholder decision-making for a Fortune 500 music industry client.
- Designed and implemented AI agents for medical professionals, following rigorous testing protocols to achieve failure rates below 10% per 100 tests, and ensured strict healthcare privacy and compliance standards.
Junior Data Scientist
Data Masters
- Assumed ownership of a critical ETL process in Microsoft SSIS, resolving an unforeseen bug on a health project within a highly regulated domain.
- Directed a small team handling data analysis, data mining, and ETL processes for a major telecommunications operator.
- Developed an XGBoost model for supply chain management at a loading facility, accurately predicting whether trucks would arrive on time or be delayed.
Member of External Relations
AIESEC
- Engaged companies to obtain sponsorships for local AIESEC events.
- Collaborated in a team, ensuring clear and effective communication.
- Reported to senior team members and managed responsibilities related to the companies I contacted through cold calling.
Experience
Predictive Maintenance Solution for Automotive Components
I successfully led the initiative to implement GNNs for the first time in an automotive ML application. I was involved from the POC stage, helping the client secure $5 million in funding. I guided the development of the model to a production-ready system with model versioning and detailed evaluation tracking.
Additionally, I developed a custom, statistically-based synthetic data generator. This approach outperformed standard artificial neural network methods and met stakeholders' performance requirements for predictive accuracy.
GraphRAG-powered AI Chatbot for eCommerce Users
http://gymshop.mkAI Agents in Healthcare
Amazon Alexa and Google Assistant App for Healthcare
Education
Bachelor's Degree in Computer Science
Faculty of Computer Science and Engineering - Skopje, North Macedonia
High School Diploma in Mathematics and Natural Sciences
Josip Broz Tito High School - Bitola, North Macedonia
Certifications
First Certificate in English (FCE)
University of Cambridge
Skills
Libraries/APIs
OpenAI API, Pandas, Scikit-learn, PyTorch, PyTorch Geometric (PyG), API Development, Claude API, XGBoost, Alexa API
Tools
Whisper, GraphRAG, Claude, Claude Code, GitHub, Claude Agent SDK, Amazon SageMaker, AWS Glue, Google Assistant, Microsoft Copilot, Grafana
Languages
Python, SQL
Frameworks
LlamaIndex, Agentic Frameworks, LangGraph
Platforms
Amazon Web Services (AWS), Jupyter Notebook, Google Cloud Platform (GCP), Vertex AI, Docker, Azure, LangSmith, AWS Lambda, WordPress, CrewAI, Azure AI Studio
Paradigms
DevOps, Automation, Event-driven Architecture, Asynchronous Programming, ETL, Synthetic Data Generation, Model Context Protocol (MCP)
Storage
Amazon S3 (AWS S3), Neo4j
Other
Data Science, Graph Neural Networks (GNNs), Machine Learning, Large Language Models (LLMs), AI Agents, Retrieval-augmented Generation (RAG), Artificial Intelligence (AI), Prompt Engineering, Solution Architecture, Predictive Modeling, Natural Language Processing (NLP), AI Design, ML Pipelines, Debugging, Model Evaluation, Multi-agent Systems, AI Pipeline, Speech-to-Text (STT), Meta Llama, AI Automation, Chatbot Conversation Design, AI Tools, APIs, Machine Learning Operations (MLOps), Perplexity, OpenAI, AI Chatbots, LangChain, Vector Databases, FastAPI, Agentic AI, Agentic RAG Systems, RAG Systems, Workflows, CI/CD Pipelines, API Integration, Cloud Architecture, Cost Reduction & Optimization (Cost-down), ETL Pipelines, CRM APIs, Workflow Automation & System Integration, AI Consulting, RAG Architecture, Cloud Platforms, System Architecture, Webhooks, Code Review, AI Model Training, Technical Leadership, Chatbots, Time Series, Time Series Analysis, AI Architecture, Vector Search, Anthropic, Data Analysis, Graphics Processing Unit (GPU), English, DVC, Linear Algebra, Algebra, Probability Theory, Statistics, Calculus, Discrete Mathematics, Competitive Programming, GPU Computing, Cold Calling, Teamwork, Communication, Data Mining, Big Data, Microsoft Data Transformation Services (now SSIS), Azure Databricks, Deep Learning, Evaluation, GraphDB, ChromaDB, Gemini, Speech Recognition, Scalable Vector Databases, Recommendation Systems, Microsoft Foundry, Statistical Modeling, Multivariate Statistical Modeling
How to Work with Toptal
Toptal matches you directly with global industry experts from our network in hours—not weeks or months.
Share your needs
Choose your talent
Start your risk-free talent trial
Top talent is in high demand.
Start hiring