Home
Hi, I’m Sabith 👋
Staff Engineer at StackGen, building enterprise AI agent platforms in Go.
I write about the engineering decisions, production bugs, and hard-won lessons from building an AI agent runtime and Aiden — a multi-tenant agent orchestration platform — at StackGen.
📚 Blog Series: Building an Enterprise AI Agent Platform in Go
-
Contributing Back While Building a Commercial Product — We built a proprietary product. We also merged 17 PRs into the agent framework we depend on. Here’s how to navigate that tension.
-
From CLI Agent to Multi-Tenant Platform — Building Aiden — A CLI tool for one developer is fun. Making it work for 50 teams with different policies is engineering.
-
Terraform for Agent Configuration — Infrastructure as Code Meets AI Governance — We use Terraform to configure our AI agents. Not YAML. Not a dashboard. Terraform. Here’s why.
-
You Can’t Debug What You Can’t See — Observability for AI Agents — Traditional APM can’t tell you why your agent spent $4.72 asking the same question three times. Here’s what agent observability actually requires.
-
Your Agent Has Root — Defense-in-Depth for AI Agents That Wield Real Tools — Your agent can run rm -rf /. Your prompt saying ‘don’t do that’ is not security. Here’s a 5-layer defense model.
-
The HITL Paradox — When Human Approval Makes Agents Worse — Human-in-the-loop is supposed to make agents safer. It can also make them useless. Here’s how to find the balance.
-
Teaching Agents to Learn Without Fine-Tuning — Post-session skill distillation from agent traces — how we teach agents to write their own runbooks.
-
Pensieve — Memory Management for AI Agents That Actually Forget — Your agent remembers everything. That’s a bug, not a feature. Here’s how we built a memory system that learns, forgets, and self-prunes.
-
Implementing ReAcTree — 6 Production Bugs the Paper Didn’t Warn You About — What happens when you take an arXiv algorithm to production. We found 6 bugs that no paper mentions.
-
52 Packages in 4 Months — Architecture at Speed Without Drowning — From Hello World to 76K lines of Go — the architecture patterns that made rapid development sustainable.
-
TOML Over YAML and PKL — How We Stopped Fighting Config and Started Shipping — We tried YAML, considered PKL, and landed on TOML for agent configuration. The reason surprised us.
-
Why We Chose Go for Our AI Agent Platform (When Everyone Else Picked Python) — Every AI framework is in Python. We built ours in Go. Here’s why we’d do it again — and when you shouldn’t.
📬 Connect
Posts
-
Contributing Back While Building a Commercial Product
-
From CLI Agent to Multi-Tenant Platform — Building Aiden
-
Terraform for Agent Configuration — Infrastructure as Code Meets AI Governance
-
You Can't Debug What You Can't See — Observability for AI Agents
-
Your Agent Has Root — Defense-in-Depth for AI Agents That Wield Real Tools
-
The HITL Paradox — When Human Approval Makes Agents Worse
-
Teaching Agents to Learn Without Fine-Tuning
-
Pensieve — Memory Management for AI Agents That Actually Forget
-
Implementing ReAcTree — 6 Production Bugs the Paper Didn't Warn You About
-
52 Packages in 4 Months — Architecture at Speed Without Drowning
-
TOML Over YAML and PKL — How We Stopped Fighting Config and Started Shipping
-
Why We Chose Go for Our AI Agent Platform (When Everyone Else Picked Python)
subscribe via RSS