Themes & categories · FWDC 2026

4 themes, 15 categories

The full detail of each theme and its categories: what we expect, and who we want to hear from. Pick the category that best matches your story; if it sits between two, pick the closest one and say so in your submission.

← Back to the Call for Speakers

Theme 01

Data Foundations for Humans & AI

Data platforms now serve two kinds of consumers: people and machines. Analysts and business teams still need trusted, well-modeled data, and a fast-growing population of LLMs and agents needs it too, with even less tolerance for ambiguity. This theme covers the foundational work that makes data consumable, trustworthy, and affordable for both.

01

AI-Ready Data

Getting data ready for machine consumers, on two fronts: clean, unambiguous data models on one side; meaning and context on the other. Agents and LLMs need both.

Topics we expect
  • Data modeling for machine consumers: source-of-truth models that minimize ambiguity at the source
  • Semantic layers and metrics stores in production
  • Ontologies and knowledge graphs for grounding AI systems
  • Context engineering for data platforms
  • Exposing governed data to LLMs and agents (including MCP-based approaches)
  • Data products designed for machine consumers

We want to hear from: data engineers, analytics engineers, data architects, and AI engineers who have made their data consumable by AI systems, and can show what it took.

02

Data Quality & Trust in the Agentic Era

When an agent reads your data and acts on it, "close enough" stops being good enough. Quality, contracts, and observability become the trust infrastructure of every AI initiative.

Topics we expect
  • Data contracts in practice: producer/consumer accountability at scale
  • Data observability that detects, explains and routes incidents
  • Trust signals exposed at decision time
  • Write-audit-publish and other safety patterns
  • Quality requirements when agents consume and act on data

We want to hear from: data engineers, platform engineers, analytics engineers, and tech leads running quality and observability in production.

03

Data Platform Evolution

The platform landscape is shifting: open table formats are becoming the default, competition is moving up to the catalog layer, and the Modern Data Stack is consolidating. Meanwhile, some teams are questioning whether distributed architectures are needed for their workloads at all.

Topics we expect
  • Open table formats and the catalog layer (Iceberg, Polaris, Unity, and friends)
  • Post-Modern-Data-Stack consolidation: migration and rationalization war stories
  • Lakehouse architectures in production
  • Single-node and small-data architectures: when distributed is over-engineering
  • Build vs buy decisions, with the trade-offs you regret

We want to hear from: data and platform engineers, data architects, and heads of data who lived through a platform decision and can share the trade-offs.

04

Cost-Aware Data Engineering

Data budgets are under scrutiny. Cost is now a design constraint, and the teams that master it ship more with less.

Topics we expect
  • Warehouse and pipeline cost optimization, with numbers
  • FinOps practices for data teams
  • State-aware orchestration and incremental processing
  • Measuring the real ROI of the data platform

We want to hear from: data engineers, engineering managers, and data leaders who cut costs without cutting capability, and can prove it.

Theme 02

Agentic Systems Engineering

Models stopped being the hard part. The engineering around them, context, memory, evaluation, orchestration, security, cost, is where AI systems succeed or fail in production. This theme is for the builders of those systems.

05

Evals & Observability

Evaluation is the production gate of AI systems: the discipline that catches regressions before your users do, and separates the few GenAI projects that ship from the many that stall.

Topics we expect
  • Eval-driven development in practice
  • LLM observability and distributed tracing
  • LLM-as-a-judge and human feedback loops
  • Regression testing for non-deterministic systems
  • Reliability engineering for AI features

We want to hear from: AI engineers, ML engineers, data scientists, and tech leads who run evals in production and can show how they changed engineering decisions.

06

Context & Memory Engineering

A large share of agent failures are context failures. Deciding what enters the context window, what gets retrieved, what gets remembered, and at what cost, has become an engineering discipline of its own.

Topics we expect
  • Context design beyond prompt engineering
  • Retrieval strategies: agentic retrieval, RAG evolutions, long-context trade-offs
  • Short-term and long-term agent memory as an infrastructure component
  • Cost / quality / latency arbitration in context design

We want to hear from: AI engineers, data scientists, and software engineers who build LLM systems and have hard-won lessons about what belongs in the context.

07

Agentic Architectures & Orchestration

Agent architectures that ship, and a clear-eyed look at those that should not have been built.

Topics we expect
  • Architectures of production agentic systems, end to end
  • When you do NOT need multi-agent systems
  • MCP and tool integration at scale
  • Orchestration patterns, failure modes, and post-mortems

We want to hear from: AI engineers, software engineers, architects, and tech leads running agents in production, especially the ones willing to share what failed.

08

Security, Sovereignty & Cost Control

Keeping control of your AI systems: who they talk to, what they can do, where they run, and what they cost.

Topics we expect
  • Prompt injection in the real world, and defenses that hold
  • Identity, least privilege, and accountability for agents
  • The MCP attack surface
  • Open-weight and European models: self-hosting as a control decision
  • Taming inference costs: model selection, caching, routing, FinOps for AI workloads

We want to hear from: AI engineers, security engineers, platform engineers, CISOs, and CTOs who secure, localize, or pay for AI systems at scale.

Theme 03

Engineering Shifts

AI is transforming engineering work itself. Whatever you build, pipelines, analytics, software, or product features, your craft is changing. This theme gives each engineering practice its own stage, with one requirement: tell us what changed for real, in a real team.

09

The Agentic SDLC

Software engineering when coding agents join the team: how specs, reviews, tests, and delivery change when part of the code is written by machines.

Topics we expect
  • Coding agents in real team workflows, beyond the solo demo
  • Spec-driven development: who writes the spec, who executes
  • Guardrails, code review, and CI for agent-generated code
  • Keeping software maintainable and extensible when agents write a growing share of the code
  • Testing, debugging, and shipping with agents in the loop
  • Honest post-mortems of what went wrong

We want to hear from: software engineers, tech leads, and engineering managers whose teams ship with agents today.

10

AI-Augmented Data Engineering

The data engineering craft, rebuilt around agents that write pipelines, fix incidents, and document as they go. What gets automated, what gets supervised, and what remains deeply human.

Topics we expect
  • Coding agents for pipelines, dbt, and migrations at scale
  • Self-healing pipelines: agents that triage, fix, and document
  • Write-audit-publish and guardrails when agents touch production data
  • What remains human in the data engineering craft

We want to hear from: data engineers, analytics engineers, and platform engineers using agents on production data systems.

11

Agentic Analytics

The next chapter of self-service: business users asking questions in natural language and getting answers they can trust, because a governed semantic layer sits underneath.

Topics we expect
  • Natural language interfaces on governed semantic layers
  • Self-service beyond dashboards
  • Agents that take action on insights, from alerting to remediation
  • What it takes to trust an AI answer: governance, lineage, evals for analytics
  • Deployment and adoption patterns, told by the teams who built them

We want to hear from: analytics engineers, data analysts, and data/BI leaders who put conversational analytics in front of real users.

12

AI Product Engineering

Shipping AI features that users actually adopt: the engineering of user-facing AI, where non-determinism meets product quality.

Topics we expect
  • Shipping user-facing AI features, from prototype to production
  • Designing for non-determinism: UX patterns for probabilistic systems
  • Product-level evals and quality metrics
  • Agentic experiences with real adoption, and how you measured it

We want to hear from: product engineers, software engineers, and AI engineers who own an AI feature in a real product.

Theme 04

Leading the AI Shift

Most Data & AI transformations stall on organizational questions: enablement, accountability, operating model. This theme is for the leaders working through them, with real numbers to share.

13

AI Enablement at Scale

How organizations move from a handful of enthusiasts to organization-wide capability: champions, academies, rituals, and a sober measurement of whether any of it works. Adoption is the visible part; the deeper work is helping people genuinely adapt, leaders included.

Topics we expect
  • AI Champions programs and internal academies: what works and what does not
  • Measuring real adoption vs declared adoption
  • From pilot to organization-wide usage: mechanics, rituals, governance of enablement
  • The ROI question: productivity claims vs measured outcomes
  • Sustaining quality of work life through the AI shift: frustrations heard, workload reality, and what your organization concretely changed
  • Leaders relearning their own job: first-hand accounts of leading a team while rebuilding your own craft

We want to hear from: data/AI leaders, engineering managers, transformation leads, and CDO/CAIO with an enablement program past the pilot stage, including the disappointing parts.

14

Governance & Accountability

Agents act, decisions get automated, and the AI Act's high-risk obligations, postponed to late 2027 by the Digital Omnibus, give organizations a window to get governance right. Governance has to answer harder questions than ever, without becoming the department of no.

Topics we expect
  • Who is accountable when an agent acts: ownership models for agentic systems
  • Shadow AI: discovery, visibility, and what to do about it
  • AI Act in practice: what already applies, what the Digital Omnibus postponed, and how to use the extra time
  • Governance that enables instead of blocking

We want to hear from: data/AI leaders, CISOs, heads of data, and compliance practitioners with concrete governance mechanisms in place.

15

Data & AI Operating Models

How the data and AI function is organized in practice in 2026: what survived the data mesh hype, where AI teams sit, and how the leadership roles are evolving.

Topics we expect
  • Centralized vs decentralized: what survived the data mesh hype
  • Hub-and-spoke and platform team patterns in practice
  • Where AI teams sit: embedded, central lab, or federated
  • The evolving CDO/CAIO role

We want to hear from: heads of data, CDO/CAIO, staff/principal engineers, and architects who reorganized a data/AI function and can compare before and after.

Found your category?

Submit your proposal through Conference Hall.

Submit my proposal →