https://aupadh12.github.io/enigmaAI/ Notes on cloud platform engineering, GitOps, and agentic infrastructure for pharmaceutical R&D — what actually works. en-us Tue, 26 Aug 2025 00:00:00 +0000 https://aupadh12.github.io/enigmaAI/blogposts/17-eks-vs-managed-runtime.html https://aupadh12.github.io/enigmaAI/blogposts/17-eks-vs-managed-runtime.html Tue, 26 Aug 2025 00:00:00 +0000 The migration looked straightforward. Then we found five categories of blockers. Here's the framework we wish we'd had before we started. Architecture https://aupadh12.github.io/enigmaAI/blogposts/16-multi-tenant-finops.html https://aupadh12.github.io/enigmaAI/blogposts/16-multi-tenant-finops.html Tue, 26 Aug 2025 00:00:00 +0000 We built the join, ran it, got 1.8% coverage. It looked like a data problem. It was a .lower() call — and two more silent bugs followed. FinOps https://aupadh12.github.io/enigmaAI/blogposts/15-gpu-pathology-ai-eks.html https://aupadh12.github.io/enigmaAI/blogposts/15-gpu-pathology-ai-eks.html Tue, 26 Aug 2025 00:00:00 +0000 Burst GPU workloads, silent Karpenter version traps, and why your SQS consumer should never be a ScaledJob. MLOps https://aupadh12.github.io/enigmaAI/blogposts/14-billion-row-data-pipeline.html https://aupadh12.github.io/enigmaAI/blogposts/14-billion-row-data-pipeline.html Mon, 18 Aug 2025 00:00:00 +0000 We had a design. The advisor caught five blockers before a single line was written. Here's what changed and why it mattered. Data Engineering https://aupadh12.github.io/enigmaAI/blogposts/13-agent-feedback-loop.html https://aupadh12.github.io/enigmaAI/blogposts/13-agent-feedback-loop.html Wed, 06 Aug 2025 00:00:00 +0000 The agent was wrong. The user corrected it. The next user got the same wrong answer. Here's how to close the loop — without fine-tuning. Learning https://aupadh12.github.io/enigmaAI/blogposts/12-cross-session-agent-memory.html https://aupadh12.github.io/enigmaAI/blogposts/12-cross-session-agent-memory.html Mon, 28 Jul 2025 00:00:00 +0000 Every session starts cold. Users re-teach the same preferences. Here's how to give your agent real memory without creating a privacy disaster. Memory https://aupadh12.github.io/enigmaAI/blogposts/11-observability-for-agents.html https://aupadh12.github.io/enigmaAI/blogposts/11-observability-for-agents.html Thu, 10 Jul 2025 00:00:00 +0000 Langfuse catches bad tool data. Prometheus catches the stuck pod. CloudWatch catches the $12 session. Here's how three layers work together — and where each one falls short alone. Observability https://aupadh12.github.io/enigmaAI/blogposts/10-teaching-agent-domain-expertise.html https://aupadh12.github.io/enigmaAI/blogposts/10-teaching-agent-domain-expertise.html Sun, 22 Jun 2025 00:00:00 +0000 Not fine-tuning. Not prompt-stuffing. A skills system that retrieves domain heuristics at query time — and the failure modes that shaped it. Domain AI https://aupadh12.github.io/enigmaAI/blogposts/9-releasing-ai-systems.html https://aupadh12.github.io/enigmaAI/blogposts/9-releasing-ai-systems.html Thu, 05 Jun 2025 00:00:00 +0000 A tag-only image bump left the UI, presentation, and diagrams five versions stale. None of it errored. Here's what a real AI system release checklist looks like. Release Engineering https://aupadh12.github.io/enigmaAI/blogposts/8-data-pipeline-for-agents.html https://aupadh12.github.io/enigmaAI/blogposts/8-data-pipeline-for-agents.html Mon, 19 May 2025 00:00:00 +0000 The agent gets the credit. The data pipeline does the work. Here's the production ingestion architecture that keeps a 19-source research agent reliable. Data Engineering https://aupadh12.github.io/enigmaAI/blogposts/7-kubernetes-to-managed-agent-runtime.html https://aupadh12.github.io/enigmaAI/blogposts/7-kubernetes-to-managed-agent-runtime.html Fri, 02 May 2025 00:00:00 +0000 I run agents on Kubernetes and evaluated moving them to a managed agent platform. Here's what genuinely improves, what quietly becomes your problem, and why a hybrid split was the honest answer. Platform Migration https://aupadh12.github.io/enigmaAI/blogposts/6-securing-agentic-ai.html https://aupadh12.github.io/enigmaAI/blogposts/6-securing-agentic-ai.html Mon, 14 Apr 2025 00:00:00 +0000 Security lessons from putting an autonomous, tool-calling agent in front of regulated enterprise data — including a cross-tenant vulnerability I shipped, found, and closed. AI Security https://aupadh12.github.io/enigmaAI/blogposts/5-benchmarking-llms-in-the-enterprise.html https://aupadh12.github.io/enigmaAI/blogposts/5-benchmarking-llms-in-the-enterprise.html Wed, 26 Mar 2025 00:00:00 +0000 Building a standing internal evaluation suite — and the four ways my harness broke before it produced a single trustworthy number. LLM Evaluation https://aupadh12.github.io/enigmaAI/blogposts/4-finops-for-agentic-ai.html https://aupadh12.github.io/enigmaAI/blogposts/4-finops-for-agentic-ai.html Sat, 08 Mar 2025 00:00:00 +0000 Your invoice tells you the total. It never tells you who spent it, which tokens were wasted, or whether the last model upgrade paid for itself. FinOps for AI https://aupadh12.github.io/enigmaAI/blogposts/3-multi-agent-collaboration.html https://aupadh12.github.io/enigmaAI/blogposts/3-multi-agent-collaboration.html Thu, 20 Feb 2025 00:00:00 +0000 Why one agent isn't enough — and how to build a team of AI specialists that research, analyze, synthesize, and critique each other's work, with full observability via Langfuse. Multi-Agent AI https://aupadh12.github.io/enigmaAI/blogposts/2-agentic-ai-systems.html https://aupadh12.github.io/enigmaAI/blogposts/2-agentic-ai-systems.html Mon, 03 Feb 2025 00:00:00 +0000 How I built a 33-tool AI agent that autonomously queries databases, builds knowledge graphs, and generates research reports — running on Kubernetes. Agentic AI https://aupadh12.github.io/enigmaAI/blogposts/1-claude-code-for-agentic-development.html https://aupadh12.github.io/enigmaAI/blogposts/1-claude-code-for-agentic-development.html Wed, 15 Jan 2025 00:00:00 +0000 How I use Claude Code's agent system to build AI systems — with a mandatory Advisor that catches production bugs, persistent memory that compounds across sessions, and a full EKS deployment architecture. AI-Assisted Development