The FDE Coach blog · new articles every day
Practical guides for engineers who ship AI.
Step-by-step builds with free AI tools, what just shipped in AI and why it matters, and Forward Deployed Engineer playbooks on interviews, pay and client work.
What just shipped in AI, and why it matters for engineers.
Page 2 of 11
AI News
GLM-5.3-Flash: What Its Price-Performance Curve Means for API Budgets
GLM-5.3-Flash disrupts the cost-quality curve. We break down the benchmarks, decode the pricing, and show engineers how to swap it into production pipelines without rewriting the architecture.
8 min read
AI News
Headlong: A Lightweight Harness for Building and Debugging Persistent AI Agents
Headlong is a bare-bones runtime that forces your AI agents to survive the long tail. Learn how this microharness simplifies state management, exposes brittle logic, and helps engineers ship reliable persistent agents.
9 min read
AI News
Building a Low-Latency AI Companion for Skyrim: Real-Time Voice, Vision, and Context
An engineer's breakdown of a real-time AI gaming companion for Skyrim that uses local LLMs, vision, and TTS. Learn the architecture, latency tricks, and why it matters for FDEs.
10 min read
AI News
AI Is Hitting Entry-Level Jobs Hardest: What the Stanford Study Means for Engineering Teams
A new Stanford study confirms AI is automating entry-level tasks the fastest. Here's how FDEs and engineering leads can redesign pipelines, mentorship, and tooling to turn this threat into a force multiplier for the whole team.
10 min read
AI News
When LLMs Attack the Host: Exploit Vectors in Inference Engine Runtimes
LLMs can now break out of inference engines to execute arbitrary code on the host. We dissect the exploit chain, why it matters for Forward Deployed Engineers, and how to reproduce it safely today.
9 min read
AI News
Apple M6 & M5 Ultra Neural Engine: What It Means for Local AI Inference
Apple's M6 and M5 Ultra pack a massive Neural Engine punch. We break down the raw specs, why it matters for local LLM inference, and how engineers can leverage this silicon shift today.
10 min read
AI News
Xiaomi’s New CPU Core: Single-Thread Parity with Apple, Multi-Thread Lead
Xiaomi's lab-built CPU core reportedly matches Apple's single-threaded performance and surpasses it in multi-threaded workloads. Here's what that means for silicon architecture, the ARM ecosystem, and the engineers building on it.
9 min read
AI News
AI Coding Is Quietly Erasing Engineering Expertise—Here's How to Stop It
AI coding assistants are accelerating output but silently eroding the deep debugging, systems thinking, and architectural intuition that define senior engineers. Here's what's happening, why it matters for FDEs, and how to build an AI-resilient skillset.
9 min read
AI News
Agentic Flooding: How Bots DDoS Government Services and How to Detect Them
A new study characterizes how LLM-powered bots overwhelm government services. We break down the attack vectors, detection signals, and what Forward Deployed Engineers can build today to harden civic infrastructure.
9 min read
AI News
The agent.md File: A Systematic Prompt to Improve LLM-Assisted Code Quality
A deep dive into the agent.md technique for taming LLM code generation. Learn how a structured markdown file can enforce coding standards, reduce hallucinations, and turn a general-purpose AI into a disciplined engineering partner.
10 min read
AI News
Your Open-Source Model Could Have a Hidden Time-Release Backdoor: How to Audit
Open-source models can conceal time-release backdoors that bypass all standard evals. Learn how these attacks work, why they threaten production pipelines, and how to audit models before they hit your inference server.
9 min read
AI News
Decayfmt: The File Format That Corrupts Itself as a Feature, Not a Bug
Decayfmt is a file format that deliberately corrupts data on every read. We break down the state-machine design, why bit-rot as a feature challenges data permanence dogma, and how Forward Deployed Engineers can use it for ephemeral prototyping.
10 min read
AI News
Claude Mythos 5 for Cybersecurity: Practical Defensive Workflows for Engineers
Anthropic's Claude Mythos 5 brings structured reasoning and massive context to defenders. Here's how engineers and FDEs can weaponize it for real-world SOC workflows, malware analysis, and threat hunting today.
9 min read
AI News
NanoGPT Speedrun Frontier: How Low Can LLM Training Costs Really Go?
The NanoGPT Speedrun smashed the cost floor for training GPT-2-class models to $4.50. We break down the algorithmic stack, why this matters for forward-deployed engineers building on the edge, and how to reproduce the results yourself.
10 min read
AI News
EU Rules AI-Generated Content Uncopyrightable: What Engineers Need to Know
The EU's stance that purely AI-generated works lack copyright protection reshapes how engineers approach code generation, tooling, and IP strategy. Here's the practical impact.
10 min read
AI News
Rust Glancer: An LSP That Uses 100x Less RAM with Tantivy Indexing
Rust Glancer slashes LSP memory from gigabytes to ~20 MB by swapping in-memory ASTs for a Tantivy-backed search index. Here's why this matters for resource-constrained editors and forward-deployed environments.
10 min read
AI News
Hacking a $27 Smartwatch with Claude Code: Embedded AI Workflows
How an engineer used Claude Code to hack a cheap smartwatch into a home automation display. The real story isn't the watch—it's the emerging workflow for embedded development where AI handles the grunt work of reverse-engineering proprietary protocols.
9 min read
AI News
Huzzah: A Novel Approach to Coding with AI That Skips the Chat Interface
Huzzah replaces prompt-chat loops with a structured task interface for AI coding. Learn why this matters for engineering velocity, how to try it, and a balanced take on its trade-offs.
10 min read
AI News
Autolith: The AI Agent That Debugs Code by Running It Live
Lambda Symbolics' Autolith gives an LLM a live Python runtime, letting it execute, observe, and fix code in a closed loop. We break down the architecture, why it matters for engineers, and how to prototype the pattern today.
11 min read
AI News
Why Your Local LLM Feels Dumber: Sampling Settings, Not Model Size
Your local LLM isn't broken—it's just badly configured. Before blaming the 7B parameter count, fix greedy sampling, broken stop tokens, and context shift. Here's how.
10 min read
AI News
Claude Code Effort A/B Tests Signal LLM-as-a-Service Pricing Shift
Anthropic is A/B testing reduced effort levels in Claude Code, triggering fewer tool calls. Here's the engineering impact on cost, latency, and automation reliability, plus how to test it today.
11 min read
AI News
Codex vs. Claude: A Week of Developer Experience Trade-offs
We swapped Claude for OpenAI Codex for a week of real engineering work. Here’s the raw log of what broke, what shipped faster, and why context-window economics matter more than benchmark scores.
10 min read
AI News
Self-Hosted Agentic Software Factory: A Blueprint for Engineers
A deep-dive into building a sandboxed, self-hosted agentic software factory using Docker, OpenHands, and local LLMs. Learn the architecture, security pitfalls, and how to build yours today.
10 min read
AI News
Sanitizing LLM Code Output: How 'Vomit' Cleans Up Claude's Token Stream
LLMs often vomit corrupted code with dangling brackets or broken syntax. Learn how a secondary 'vomit' agent cleans the token stream, why this matters for production pipelines, and how to implement it today.
9 min read
259 articles and counting