AI engineer, consultant, and founder. Writing about LLMs, distributed systems, and building things that work.https://prabal.ca/https://prabal.ca/posts/google-long-context-cheaper/https://prabal.ca/posts/google-long-context-cheaper/Google's TurboQuant compresses the KV cache by ~5x with minimal quality loss - what this means for open-source LLM inference.Thu, 26 Mar 2026 14:38:23 GMThttps://prabal.ca/posts/monitorintent-ai-outreach/https://prabal.ca/posts/monitorintent-ai-outreach/The first product at MonitorIntent was clever - maybe too clever. AI messages that sounded human but made judgment errors nobody forgave.Sat, 13 Sep 2025 00:00:00 GMThttps://prabal.ca/posts/monitorintent-building-mvp/https://prabal.ca/posts/monitorintent-building-mvp/Building MonitorIntent's MVP - 2 months to a Replit dashboard, 5x LLM cost reduction, and the realization that LinkedIn likes are not buying intent.Sat, 20 Sep 2025 00:00:00 GMThttps://prabal.ca/posts/monitorintent-postmortem/https://prabal.ca/posts/monitorintent-postmortem/The post-mortem: 5 things that killed a bootstrapped B2B startup, what's salvageable, and what I'd do differently.Sat, 04 Oct 2025 00:00:00 GMThttps://prabal.ca/posts/monitorintent-selling-dying/https://prabal.ca/posts/monitorintent-selling-dying/Selling MonitorIntent to startup founders, the Clay defection, the vitamin problem, and watching our moat disappear in real time.Sat, 27 Sep 2025 00:00:00 GMThttps://prabal.ca/posts/x-algorithm-visualizer/https://prabal.ca/posts/x-algorithm-visualizer/An RPG-style simulation of X's recommendation algorithm - pick a persona, write a tweet, and watch the pipeline decide if it goes viral. Real ML running in the browser.Sun, 25 Jan 2026 00:00:00 GMThttps://prabal.ca/posts/latentscore-research/https://prabal.ca/posts/latentscore-research/An open-source Python library that turns text prompts into procedural music in real time, on CPU, sub-second.Mon, 02 Mar 2026 00:00:00 GMThttps://prabal.ca/posts/llm-control-production/https://prabal.ca/posts/llm-control-production/Making LLMs produce reliable JSON - Part 3 of 3. Temperature tuning, graceful degradation, observability, token caching, jailbreak detection.Mon, 24 Nov 2025 00:00:00 GMThttps://prabal.ca/posts/llm-control-multi-model/https://prabal.ca/posts/llm-control-multi-model/Making LLMs produce reliable JSON - Part 2 of 3. Multi-model architectures, writer-critic loops, best-of-N sampling, and the freeform thinker + JSON formatter pattern.Mon, 17 Nov 2025 00:00:00 GMThttps://prabal.ca/posts/llm-control-schema-design/https://prabal.ca/posts/llm-control-schema-design/Making LLMs produce reliable JSON - Part 1 of 3. Simple schemas, detailed descriptions, constrained decoding, and why enums beat strings.Mon, 10 Nov 2025 00:00:00 GMThttps://prabal.ca/posts/claude-code-chatgpt-subscription/https://prabal.ca/posts/claude-code-chatgpt-subscription/Use your existing ChatGPT Plus/Pro/Max quota to power Claude Code via LiteLLM - no separate API key needed. Now with background launchd service and silent startup.Wed, 01 Apr 2026 22:37:58 GMT