This is a channel about agentic engineering — the practical craft of building AI agents and LLM integrations that survive production and don't fall apart after the first week.
I'm Walter (@balialin). I build on Claude/OpenAI: Telegram bots, scraping pipelines with LLM extraction, agents with tool use, MCP servers. Currently in production: an equity screener with LLM Web Search, an App Store review analyzer with Word/Excel reports, and a couple of bots saving operators dozens of hours a week.
What I'll write about: things that are actually hard. Rate limits. Caching strategies. The errors that break agents in production. Token economics. MCP patterns. When Claude Agent SDK genuinely beats "I'll just write my own agent."
What I won't write: "10 best AI tools," "make a million on prompts," any flavor of hype-merchant content.
Roughly one post per week. More often when something is on fire.
Site: balialin.dev
I'm Walter (@balialin). I build on Claude/OpenAI: Telegram bots, scraping pipelines with LLM extraction, agents with tool use, MCP servers. Currently in production: an equity screener with LLM Web Search, an App Store review analyzer with Word/Excel reports, and a couple of bots saving operators dozens of hours a week.
What I'll write about: things that are actually hard. Rate limits. Caching strategies. The errors that break agents in production. Token economics. MCP patterns. When Claude Agent SDK genuinely beats "I'll just write my own agent."
What I won't write: "10 best AI tools," "make a million on prompts," any flavor of hype-merchant content.
Roughly one post per week. More often when something is on fire.
Site: balialin.dev