Dejan
Petković

Elands AI designs and delivers AI platforms for startups and small companies.

• Designing and shipping LLM-based applications for clients
• Building evaluation infrastructure for LLM systems
• Building real-time AI services

I joined Delphyr to build an AI platform specifically for the medical field. We are bringing AI into real clinical workflows, from hospitals to GP practices.

• We won our first contract with a Dutch mental health provider, by successfully completing integration with an EHR. On top of that I lead EHR integrations across four vendors
• I work with several coding agents in parallel. Each runs on a plan from Linear and a second model reviews its work
• I led the effort to encrypt data with KEK/DEK envelope encryption and to iterate on pentests to harden the system. I automated the UDI-PI release process
• With the product platform in place, I now focus on the AI layer. This consists of prompt design for various workflows, guardrails, evals, integrations and the observability that keeps model behavior measurable in production
• I built release gating on measured eval regressions and cut the eval suite runtime from 55 to 25 minutes. LangWatch observability with PHI-safe tracing shows where the chat harness needs improvement
• I ran embedding migration from Azure to Nebius, first evaluating and then implementing the migration. I migrated inference providers (Nebius, Baseten) and enabled visibility into token spending
• I worked on prompt design and grounding for ambient clinical documentation, from per-specialism templates to transcript-level verification of every generated statement
• I set up the Nebius Kubernetes cluster, ArgoCD-based GitOps and self-hosted CI runners. Fully automated tag-driven releases run evals and integration tests to validate deployment
• I lead incident command and write postmortem analysis for our production incidents

Development of a greenfield project with a new tech stack, from scratch.

• Event-driven architecture was used to capture gift card events and process them in real-time
• I worked across the stack in a team of three engineers, a designer, and a product owner
• Management of App Store and Play Store releases

Team lead managing a team of 5 backend and frontend developers on grndhouse.com, a platform for on-demand strength training.

• Led the development of the initial MVP after a successful funding round
• Met stakeholders regularly to plan the video streaming, payment processing, and user management features
• Designed and built features with React Native, Node.js, and AWS
• Mentored junior developers and cleared blockers for the team

Progressed from individual contributor to team lead, managing a team of 5 backend and frontend developers on fertifa.com, a reproductive healthcare provider.

• Defined and developed features in collaboration with the product owner and stakeholders
• Built features in React Native, Node.js, and AWS

Senior engineer in the team developing an ETL automation framework and platform in Azure, based on the Hadoop & Kafka ecosystem, automated by Terraform.

• Rendering framework built with Python, Flask and SQLAlchemy
• Led story mapping for migration of on-premise data platform services to Azure
• Development of a distributed event-sourcing application for real-time analysis of the KPN customer journey, where multiple touch points are combined, analysed and visualised
• Scala stack with Akka / Alpakka used to build microservices and Flink applications
• DevOps methodology with fully automated CI/CD, monitoring and alerting

As technical lead of a 3-member agile team I was responsible for development of human-machine interaction on the artificial intelligence framework Amelia.

• Mostly hands-on development, code reviewing, mentoring and Scrum rituals
• Development mostly in Groovy / Python and in Java
• Data analysis with Pandas and Jupyter, visualisation with d3.js, and making new feature (or refactoring) decisions from the insights

As member of Liberty Global's OTT streaming solution architecture & engineering team, I worked on the architecture of various system components, developing AWS (micro)services infrastructure, CDN content delivery, Adobe AEM and their integration.

• Microservices development in Python
• Deep knowledge of web and backend application design, streaming technologies, HA scalable infrastructure, AWS cloud and full stack network protocols, acquired through work with high availability services
• Data analysis with Jupyter, Wireshark, Conviva and Omniture
• Performing code reviews and overall system troubleshooting