Uploaded June 2026 | Updated November 2026, 19 hours ago
Debugging is where real engineering happens, and WebAssembly is reaching the point where “you can instrument it” isn’t good enough. In this episode of WebAssembly Unleashed, Joel Moses and Oscar Spencer sit down with Elizabeth Gilbert, a PhD candidate at Carnegie Mellon, to talk about Whamm: a DSL and framework designed to make instrumentation in Wasm less of a one-off engineering project and more of a repeatable, portable capability.
Elizabeth explains how she arrived at WebAssembly through the developer tooling and fault-injection world, and why existing instrumentation approaches in Wasm often boil down to fragile bytecode rewriting. Whamm reframes the problem as “match and inject,” letting you express instrumentation at a higher level and then compile the probe logic as WebAssembly itself. That means you can write the instrumentation in any language that compiles to Wasm, while Whamm handles the hard parts: matching at opcodes, function entry/exit, branches and branch targets, imports, and other low-level control-flow details that are easy to get wrong.
The conversation also covers an important evolution: running Whamm through an engine interface instead of pure rewriting. In that model, the runtime understands probes as probes, giving stronger isolation guarantees and opening the door to a standardized, portable instrumentation mechanism across engines.
Finally, Elizabeth previews Splicer, her component-model work that can interpose on edges in a component composition graph and visualize those relationships, plus a candid look at where things get hard, especially around resources.
If you care about observability, debugging, security analysis, or performance tooling for WebAssembly, this is a clear look at why the future may be less about more tools and more about better ways to describe instrumentation.
Chapters:
00:00 Welcome to WebAssembly Unleashed
00:43 WASI Preview 3 Update
01:28 Why debugging is the real engineering (and why Whamm exists)
03:06 Developer tooling led to Wasm: Distributed systems + CI/CD pain
04:18 Whamm in one line: A DSL for match-and-inject instrumentation
07:53 What Whamm can instrument: Opcodes, calls, branches, entry/exit
11:26 Keeping tooling from wrecking production: Sandboxed probes
12:45 Two modes: Bytecode rewriting vs engine-level instrumentation
15:14 Monkey patching with “alternate probes” (yes, it’s possible)
16:40 Why it’s called Whamm: “Put some whamm on it”
17:51 Splicer preview: Component-model instrumentation + interposing on graphs
26:28 Is DSL the answer to WebAssembly debuggability and observability?
28:33 Component model's hardest part: Resources + adapters (not just primitives)
For more from F5's Office of the CTO visit the following sites:
Blogs - f5.com/company/octo
Reports - f5.com/services/resources/reports
Meet Your Hosts:
Joel Moses | linkedin.com/in/joelmoses | community.f5.com/users/joel_moses/398372
Oscar Spencer | twitter.com/oscar_spen | linkedin.com/in/oscarspen
Matthew Yacobucci | linkedin.com/in/matthew-yacobucci-323b4b2
⬇️⬇️⬇️ JOIN THE COMMUNITY! ⬇️⬇️⬇️
DevCentral is an online community of technical peers dedicated to learning, exchanging ideas, and solving problems - together.
Find all our platform links ⬇️ and follow our Community Evangelists! 👋
➡️ DEVCENTRAL: community.f5.com
➡️ YOUTUBE: youtube.com/devcentral
➡️ LINKEDIN: linkedin.com/showcase/f5-devcentral
➡️ TWITTER: twitter.com/devcentral
Your Community Evangelists:
👋 Jason Rahm: linkedin.com/in/jrahm | twitter.com/jasonrahm
👋 Buu Lam: linkedin.com/in/buulam | twitter.com/buulam
👋 Aubrey King: linkedin.com/in/aubreyking | twitter.com/aubreykingf5
👋 Chase Abbott: linkedin.com/in/chaseabbott1
Debugging is where real engineering happens, and WebAssembly is reaching the point where “you can instrument it” isn’t good enough. In this episode of WebAssembly Unleashed, Joel Moses and Oscar Spencer sit down with Elizabeth Gilbert, a PhD candidate at Carnegie Mellon, to talk about Whamm: a DSL and framework designed to make instrumentation in Wasm less of a one-off engineering project and more of a repeatable, portable capability.
Elizabeth explains how she arrived at WebAssembly through the developer tooling and fault-injection world, and why existing instrumentation approaches in Wasm often boil down to fragile bytecode rewriting. Whamm reframes the problem as “match and inject,” letting you express instrumentation at a higher level and then compile the probe logic as WebAssembly itself. That means you can write the instrumentation in any language that compiles to Wasm, while Whamm handles the hard parts: matching at opcodes, function entry/exit, branches and branch targets, imports, and other low-level control-flow details that are easy to get wrong.
The conversation also covers an important evolution: running Whamm through an engine interface instead of pure rewriting. In that model, the runtime understands probes as probes, giving stronger isolation guarantees and opening the door to a standardized, portable instrumentation mechanism across engines.
Finally, Elizabeth previews Splicer, her component-model work that can interpose on edges in a component composition graph and visualize those relationships, plus a candid look at where things get hard, especially around resources.
If you care about observability, debugging, security analysis, or performance tooling for WebAssembly, this is a clear look at why the future may be less about more tools and more about better ways to describe instrumentation.
Chapters:
00:00 Welcome to WebAssembly Unleashed
00:43 WASI Preview 3 Update
01:28 Why debugging is the real engineering (and why Whamm exists)
03:06 Developer tooling led to Wasm: Distributed systems + CI/CD pain
04:18 Whamm in one line: A DSL for match-and-inject instrumentation
07:53 What Whamm can instrument: Opcodes, calls, branches, entry/exit
11:26 Keeping tooling from wrecking production: Sandboxed probes
12:45 Two modes: Bytecode rewriting vs engine-level instrumentation
15:14 Monkey patching with “alternate probes” (yes, it’s possible)
16:40 Why it’s called Whamm: “Put some whamm on it”
17:51 Splicer preview: Component-model instrumentation + interposing on graphs
26:28 Is DSL the answer to WebAssembly debuggability and observability?
28:33 Component model's hardest part: Resources + adapters (not just primitives)
For more from F5's Office of the CTO visit the following sites:
Blogs - f5.com/company/octo
Reports - f5.com/services/resources/reports
Meet Your Hosts:
Joel Moses | linkedin.com/in/joelmoses | community.f5.com/users/joel_moses/398372
Oscar Spencer | twitter.com/oscar_spen | linkedin.com/in/oscarspen
Matthew Yacobucci | linkedin.com/in/matthew-yacobucci-323b4b2
⬇️⬇️⬇️ JOIN THE COMMUNITY! ⬇️⬇️⬇️
DevCentral is an online community of technical peers dedicated to learning, exchanging ideas, and solving problems - together.
Find all our platform links ⬇️ and follow our Community Evangelists! 👋
➡️ DEVCENTRAL: community.f5.com
➡️ YOUTUBE: youtube.com/devcentral
➡️ LINKEDIN: linkedin.com/showcase/f5-devcentral
➡️ TWITTER: twitter.com/devcentral
Your Community Evangelists:
👋 Jason Rahm: linkedin.com/in/jrahm | twitter.com/jasonrahm
👋 Buu Lam: linkedin.com/in/buulam | twitter.com/buulam
👋 Aubrey King: linkedin.com/in/aubreyking | twitter.com/aubreykingf5
👋 Chase Abbott: linkedin.com/in/chaseabbott1




![Optimizing AI Inference using NGINX Gateway Fabric
Running AI inference workloads on Kubernetes presents unique challenges. Traditional load balancing methods designed for standard web APIs simply fall short when it comes to the nuanced resource management required for generative LLMs.
In this demo, we explore how the NGINX Gateway Fabric Inference Extension transforms your Kubernetes environment by acting as a dedicated Inference Gateway. Discover how to move beyond basic proxying by utilizing a powerful two-stage architecture—combining standard HTTPRoutes with the InferencePool CRD and Endpoint Picker. This allows NGINX to make intelligent, real-time routing decisions based on queue depth and GPU availability.
What you will see in this demo:
0:11 - The Challenges of AI Workloads: Why standard load balancing fails for GPUs.
0.51 - Two-Stage Orchestration: Understanding the InferencePool and Endpoint Picker.
2:24 - Model-Aware Routing: Inspecting traffic paths to offload lightweight tasks to CPUs and reserve premium GPUs for heavy text generation.
3:46 - Canary Testing: Validating new models safely using native 90/10 traffic splitting.
4:49 - Cost Optimization: Using custom HTTP headers to guarantee SLAs and reserve compute for mission-critical queries.
🔗 Helpful Links & Resources:
NGINX Gateway Fabric Documentation: [https://docs.nginx.com/nginx-gateway-fabric]
GitHub Repository / Lab Guides: [https://github.com/nginx/nginx-gateway-fabric]
Notes:
- Contributed by: Akash Ananthanarayanan
- Related Article: None
⬇️⬇️ JOIN THE COMMUNITY! ⬇️⬇️⬇️
DevCentral is an online community of technical peers dedicated to learning, exchanging ideas, and solving problems—together.
Find all our platform links ⬇️ and follow our Community Evangelists! 👋
➡️ DEVCENTRAL: https://community.f5.com
➡️ YOUTUBE: https://youtube.com/devcentral
➡️ LINKEDIN: https://www.linkedin.com/showcase/f5-devcentral/
➡️ X: https://x.com/devcentral
Your Community Evangelists:
👋 Jason Rahm: https://www.linkedin.com/in/jrahm/ | https://x.com/jasonrahm
👋 Buu Lam: https://www.linkedin.com/in/buulam/ | https://x.com/buulam
👋 Chase Abbott: https://www.linkedin.com/in/chaseabbott1 Optimizing AI Inference using NGINX Gateway Fabric](https://i.ytimg.com/vi/TReqi7V8yoI/mqdefault.jpg)





