Unlocked is dedicated to the infra nerds.
We're bringing together experts who build and operate web-scale infrastructure to share real-world lessons about real-world architectures with Valkey and Redis.
If you’re obsessed with distributed systems, tail latency, sub-millisecond response times, and evolving systems engineering for agents - this is where you need to be!
HEAR FROM INDUSTRY LEADERS:
Speakers
Agenda
Breakfast

Opening Remarks
Kick-start the day with a foreword from Khawaja Shams and Mas Kubo

The Power of Community
How community-driven collaboration is accelerating Valkey innovation and shaping the future of modern infrastructure.

The 3AM Test: Why Boring Systems Let You Sleep At Night
How Valkey is evolving Day 2 operations for predictable systems at scale

Curating the Open Source Inbox with Agents and Valkey
What happens when your software can perform its own routine operations leaving you free to do whatever you want?

Lessons from Building a Production-grade Valkey Client Library in Swift
Building clients that survive worst-case production scenarios
Break

Valkey and Semantic Caching
Stop paying for repeated LLM calls when queries mean the same thing

Tweaking Valkey for High Write Rate Workloads
How to handle replication lag, memory pressure, and contention at high write rates

Beyond fork(): Memory-Efficient Snapshots for Valkey
Stop doubling memory just to take a snapshot

Finding Hidden Headroom in Saturated Clusters
Why 100% CPU doesn’t mean your cluster is out of capacity
Lunch

Geo-Replication with Valkey
How Apple keeps Valkey clusters in sync across regions without losing data

Scaling Search with Multithreading and Hybrid Queries
Combining vector, text, and filters in a single high-performance query

Foundations for High-Throughput Storage Envoy Plugins
When your proxy becomes the bottleneck and how to fix it

Efficiency at Scale: Our Journey from Redis to Valkey
Optimizing access patterns and memory to scale beyond Redis limits
Break

Secure, Scalable TLS Enhancements to Harden Valkey
Automatic TLS rotation, client auth, and guardrails without downtime

Towards Faster Inference: With KV Cache and Beyond
Modern LLM inference is a caching problem in disguise and the instincts that built your last cache will only get you halfway there.

Not All RESP Clients Are Created Equal
What to do when you can't rely on connection pooling at scale

Thinking beyond demand filled caching: how versioned caches dominate Netflix
How Netflix uses versioned caches to rethink the hardest problems in caching
Afternoon Break

Stress-Testing Valkey with the "Valkey Lab" Benchmarking Tool
Measure real performance gains with workloads that match production

Solving Cache Drift in High-Stakes Financial Systems
How to make the right trade-offs in financial production systems.

The 3AM Page: Tales From The Trenches
Raw, unscripted production horror stories and how to prevent them
Happy Hour!






.jpg)

.jpeg)

.png)

.png)

.jpg)

.jpg)
.jpeg)

.jpeg)
.jpg)

.jpg)

.jpg)
.jpg)
.jpeg)
.jpeg)
.png)

.jpg)
.jpg)

.jpeg)

.jpg)
.jpeg)

.jpeg)
.jpg)
.png)
.jpeg)

.jpg)
.jpg)