Dragutin Oreški

Timeline

May 2026
Migrated a WhatsApp health chatbot off Infobip Answers to Python with Claude Code’s /goal and dynamic workflows — written up on the blog.
Apr 2026
Created UI Suputnik as a companion for my Algebra Bernays students, helping them revisit AI concepts and practice between lectures.
Mar 2026
Set up reusable skills that automate ETL work we previously had to do by hand — turning recurring data wrangling into one-shot agent calls.
Jan – Feb 2026
Implemented company-wide Claude Code usage policies and ran upskilling sessions so the team can lean on it safely and effectively.
Dec 2025
Diving into deep agent patterns across product and data — mapping where multi-step planning actually pays off.
Oct – Nov 2025
Exploring LLM and agent observability — finding the right traces, evals, and dashboards to actually trust what ships.
Aug – Sep 2025
Leveraging the latest Context Engineering trends via LangGraph to keep agent state lean and predictable.
Jun – Jul 2025
Optimizing self-hosted Llama models through speculative decoding and aggressive caching — chasing real cost wins.
Mar – May 2025
Building out the first private RAG platform for secure document chat and agentic workflows on customer data.

Work

Jun 2026

MEGI Health
Founding Engineer
Building AI for proactive maternal and cardiovascular care.
- Migrated the WhatsApp delivery layer off Infobip Answers into owned, test-covered Python — the write-up.
- Working on LLM-driven patient conversations bounded by deterministic clinical rules.
Apr 2026

Algebra Bernays
Lecturer · 60-hour AI benefits and risks course
UI Suputnik course companion
Lecturing on practical AI foundations for business decision-making.
- Teaching AI, machine learning, and deep learning concepts through guided demos and practical exercises.
- Connecting data preparation, model workflows, and open-source tools with real business use cases.
- Covering risks, ethics, and responsible adoption alongside the technical foundations.
Dec 2024–Jun 2026

QED Code
Lead AI Engineer
Worked on SEERai by Galorath Incorporated through QED Code.
- Focused exclusively on SEERai by Galorath Incorporated, which was named 2025 Global AI Awards Best of Category in Agentic AI.
- Designed a LangGraph multi-agent system for generating project breakdowns from private customer documents.
- Developed a private RAG platform for secure document chat and agentic workflows.
- Optimized self-hosting of Llama models for high performance and cost efficiency.
- Implemented Langfuse for observability and DeepEval for evaluation.
Mar 2024–Dec 2024

Freelancer
AI Engineer · Top 3% on Upwork
Built custom RAG and multi-agent systems for a range of clients.
- Built AI chatbots:
  - Storyteller — AI-generated personalized children’s stories.
  - TeslAI / Chat with Nikola — interactive Nikola Tesla chatbot.
  - AI Therapist — CBT-driven anxiety support.
- Designed an AI-based product matching system.
- Developed a simulation engine letting game providers test and optimize betting parameters via outcome forecasting.
- Designed and deployed conversational AI solutions using Dialogflow.
- Built a scalable video creation tool for custom videos and large-scale greenscreen overlays.
Sep 2023–Mar 2024

Upp Agency
Co-founder
Co-founded an agency focused on AI & Data Engineering.
- Led a team of 5 engineers maintaining and improving data pipelines and dashboards.
- Designed and migrated infrastructure to dbt, Snowflake, Prefect, and Kubernetes.
Aug 2017–Sep 2023

CIAL Dun & Bradstreet
Data Engineer / Scientist
Designed distributed ELT pipelines and NLP solutions for Latin-American markets.
- Developed automated data processing pipelines and analytics for business use cases.
- Implemented NLP-based entity and company matching solutions.
- Mentored new hires and established team processes.