02 / Engineering Stage Oct 15–16 · Sarajevo Bosnian Cultural Center

You already run
their code.

The people who build Cursor, vLLM, Hugging Face and NVIDIA's stack are on one stage for two days, with the teams running it where downtime costs real money. Twenty-minute technical sessions. Architectures, numbers, failure modes.

Get your Engineering Stage ticket Get your ticket
Check your stack against the speaker list Check the speaker list
18+companies
20minute talks
02technical days
0sales decks
THE STACK
IN PRODUCTION

Models built at Google. Open-source tooling from Hugging Face. NVIDIA silicon beneath vLLM and Red Hat infrastructure. Data in a Databricks lakehouse or Supabase Postgres. Vectors and logs in Elastic. Code written in Cursor, merged through GitHub, built and tested on remote-execution infrastructure like NativeLink. Voice through ElevenLabs.

You picked these tools from docs, benchmarks, and changelogs. You debugged them through GitHub issues at 2 a.m. On October 15–16, the people who build and run this stack are in the same room, taking questions.

Next to them: teams running AI where failure costs real money — across aircraft maintenance, mobility, industrial equipment, telecom infrastructure, and software used by businesses at scale.

Twenty-minute technical sessions. Architectures, numbers, and failure modes. What it costs to run, where it broke, and what they would build differently today. If a talk works as an ad, it does not make this stage.

Published lineup

The engineers behind the stack.

18 speakers and counting

The programme

The agenda.

Bosnian Cultural Center · Main Hall

Full summit agenda
  1. – 10:20
    opening

    Stage Opening

  2. – 10:50
    keynote

    What Remains When Code Writes Itself?

    About this keynote

    As generative AI lowers the cost of writing code to zero, the traditional software engineering paradigm is undergoing a massive shift. Syntax is now a commodity but faster code generation does not eliminate failure; it merely shifts the bottleneck. When code writes itself, what remains? The answer is data and enterprise context. An AI can write a flawless query in milliseconds, but without a deep, governed understanding of your organization’s unique domain definitions such as how your business measures profit margins, customer satisfaction, or employee turnover it simply produces high-speed noise. This keynote explores why the true enterprise value is no longer the codebase itself, but the underlying data foundation and the semantic context that explains what the data actually means. Discover how software builders are transitioning from writing code to curating business truth and domain logic. We will examine how Databricks Genie and Genie Ontology operationalize this shift by anchoring AI agents in unified Unity Catalog semantics and learned context, transforming raw schemas into accurate, deterministic, and actionable business intelligence.

  3. – 11:20
    keynote

    Overloaded: The Human Brain, New Jobs and the Real Cost of AI

  4. – 11:50
    keynote

    AI in Enterprise

  5. – 12:40
    workshop

    What the Model did not Learn: Sparse Multimodal Vector Search

    About this keynote

    A model’s training ends, but its usefulness doesn’t. This talk follows my attempt to turn a frozen dense embedding model, jina-embeddings-v4, into sparse vectors that a search engine’s inverted index could use, sparked by a paper called “No More K-means.” It’s really two stories: the research itself, and what doing it with a coding agent taught me. Expect failed experiments, an unexpected comeback, and practical lessons.

  6. – 13:00
    keynote

    Building Digital Future Through Partnership and Cooperation

  7. – 14:00
    break

    Networking break powered by Ptrade Lavazza

  8. – 14:30
    Fireside chat

    From Sarajevo to Zagreb — How Regional Tech Is Powering the AI Telco

  9. – 15:10
    panel

    Builders and Operators — What It Takes to Scale AI in the Enterprise

    Moderated by Džefris Dizdarević · Lufthansa Technik

    Supported by GIZ Bosnia and Herzegovina

  10. – 15:45
    workshop

    Your App Needs an MCP Server

  11. – 16:20
    workshop

    Physical AI, Open to All: Building with Hugging Face LeRobot

  12. closing

    Closing remarks

Companies on this stage

The organisations shaping the AI stack.

Buy tickets

Bring the problem you're stuck on. Someone in this room has already paid for the answer.

Date 15–16 October 2026
Venue Bosnian Cultural Center
Format Two full days
Focus From prototype to production

Inside the Engineering Stage

One technical journey.
Three layers of work.

These are programme territories, not a published timetable. Session times and speaker pairings will appear only after they are confirmed.

01 BUILD

Models, agents & interfaces

Start where the product becomes intelligent: model behavior, agent orchestration, voice, and open systems.

  1. 01

    Speculative Decoding

  2. 02

    Building AI Agents — Multi-Agent Systems and Orchestration

  3. 03

    What the Model Forgot to Learn

  4. 04

    Giving AI a Voice — Building with Real-Time Speech Models

  5. 05

    Open-Source AI in Practice

02 SHIP

Data & production reliability

Move beyond the demo: retrieval, owned data, right-sized models, and systems that hold up when failure matters.

  1. 01

    RAG in Production — Retrieval Done Right

  2. 02

    Lakehouse to LLM — Building AI on Your Own Data

  3. 03

    Why Bigger AI Isn't Always Better

  4. 04

    AI in Mission-Critical Engineering

03 SCALE

Platforms & the software factory

Connect enterprise architecture with the AI-native workflows changing how software is planned, built, and maintained.

  1. 01

    Enterprise AI — Architecture, Integration, and Scale

  2. 02

    From Idea to Production — Building with Supabase

  3. 03

    Shipping Faster with GitHub Copilot and AI-Native Workflows

  4. 04

    Building Your Own Software Factory with Cursor

  5. 05

    Building Software That Builds Itself

Programme proposal

The technical layer worth pushing further.

Four additions that close the gap between building impressive demos and operating dependable AI systems.

01

Agent Evaluation & Observability

Failure testing, traces, quality gates, and production feedback loops.

02

Agent Security

Prompt injection, tool permissions, memory poisoning, and human approval.

03

The Interoperable Agent Stack

MCP, A2A, and the architecture of agents that work across systems.

04

Inference Economics

Latency, quantization, small models, and on-device AI that can actually ship.

15–16 October 2026 · Sarajevo

Build the future.
Then ship it.

Subscribe to Our Newsletter

Elevate your understanding of emerging trends, gain access to exclusive insights and stay at the forefront of innovation by subscribing to our newsletter.

Kiss the future

/ Gajev Trg 4, 71000 Sarajevo, BiH

/ E-mail: info@bluminstitut.ai

/ Call: +38733821822

© 2026 Bloomteq. All Rights Reserved.

Privacy Policy