AI Engineer Paris is back! This year join 1000+ attendees for a community-driven conference, inspired and supported by the team behind the AI Engineer World’s Fair.

This two-day, four-track technical conference and expo brings together AI engineers, CTOs, and VPs of AI to connect, learn, and engineer the future of AI. Full details are available at ai.engineer/paris

We're looking for technical, practitioner-driven talks across the following focus areas. Submissions should favor real architectures, hard-won lessons, and shipping experience over demos and hype.

Talks are 25 minutes long, and live demos are welcome.

Key Dates:

• July 31: Submission deadline
• August 14: All proposals notified
• August 26: Speaker info session (must be available to attend)
• September 23: Rehearsals for Main Stage
• September 23: Speaker Dinner
• September 23: Welcome Reception & Opening Keynote
• September 24: Full-Day Conference featuring keynote talks, workshops, fireside chats, and after party

Submit  Your Proposal

Due to the high volume of proposals we receive, please include a link to your prior speaking experience for evaluation.

Choose from one of the following topic areas:

Agentic Engineering & Agents in Production

Designing, orchestrating, and operating agents that take real actions in production. Talks can cover single- and multi-agent architectures, tool use and function calling, planning and control loops, human-in-the-loop patterns, and the reliability, governance, and cost challenges of running agents at scale.

Context & Harness Engineering

Everything around the model that makes it perform. Topics include context-window management, retrieval, memory, compaction and caching, prompt and instruction design, and the scaffolding ("harness") that turns a raw model into a dependable system. We're looking for deep, practical techniques rather than introductions.

Coding Agents & Software Factories

How AI is reshaping the software development lifecycle, from code completion to autonomous coding agents and "software factory" workflows. Topics include code generation and review, spec-driven development, repo-scale context, agent integration into CI/CD, and how teams actually measure developer productivity.

Evals & Observability

The discipline of knowing whether your AI system actually works. Talks can cover evaluation frameworks and benchmarks, scoring probabilistic and multi-turn outputs, regression testing, tracing and monitoring in production, and building the feedback loops that catch failures before users do.

Open, Local & Sovereign AI

Running AI on your own terms: open-weight models, self-hosting and on-prem/edge deployment, model-switching and avoiding vendor lock-in, and the data-residency, privacy, and compliance realities that matter in Europe and in regulated environments. We especially welcome concrete deployment stories.

Document AI, Vision & OCR

Turning unstructured documents and images into structured, usable data. Topics include OCR and layout understanding, multimodal and vision-language models, document Q&A and extraction pipelines, and high-accuracy processing for complex, multilingual, or regulated document workflows.

Inference & AI Infrastructure

The systems layer beneath every AI product: serving and throughput, quantization, KV-cache and batching, multi-GPU and cost/latency optimization, sandboxing, and the platform engineering that keeps production AI fast and affordable.

Enterprise & Vertical AI in Regulated Industries

Real deployments in finance, healthcare, public sector, and industry, where governance, security, and compliance are first-class constraints. Talks should share concrete architectures, adoption lessons, risk and audit approaches, and measurable business outcomes.

Voice & Realtime AI

Building responsive voice and real-time multimodal experiences: speech-to-text and text-to-speech, voice agents, streaming and latency management, turn-taking, and the design challenges of conversational interfaces that feel natural.

Fine-tuning, Post-training & Custom Models

Adapting models to your domain and data. Topics include fine-tuning and post-training techniques, RL and preference optimization, building and curating high-quality datasets and RL environments, and the trade-offs between prompting, retrieval, and training a custom model.