# Blog
Notes on applied AI engineering — document intelligence, evaluation, and systems that survive contact with real data.
16 POSTS
2026
16SLMs for Swiss Re: Supervised, Structured, and Serving
For insurers, the winning pattern is rarely “largest model everywhere.” It is smaller, supervised, schema-bound models embedded into claims, underwriting, and compliance workflows with humans still controlling the decision.
ParaEval and the CRP/HDP Model: Bayesian Nonparametric Trigger Calibration for Parametric Insurance
Parametric insurance triggers are only , which uses Bayesian nonparametric clustering to discover latent peril regimes and calibrate triggers with lower basis risk than standard actuarial baselines.
Your Resume Is Also a Prompt: Why Prompt Injection Is the Defining Security Problem in Real-World LLM Systems
The most dangerous misunderstanding in enterprise AI is treating documents , connects it to broader attack surfaces, and argues that interface design is now security design.
Security & Compliance Standards for AI Systems
AI security begins where ordinary app security stops: the attack can be a dataset, a gradient, or a paragraph that looks harmless. This guide maps that wider threat surface and the controls regulated teams need.
Operating AI in Regulated Environments: HIPAA, GDPR, PCI DSS & Beyond
The moment an AI system touches health, payment, or EU personal data, architecture turns into compliance choreography. This guide translates the major regulations into the engineering artifacts and process controls they demand.
Parametric Insurance Deep Dive: Speed, Basis Risk, and Trigger Design
Parametric insurance replaces loss adjustment with an observable trigger, but the hard part is not speed. It is picking a defensible index, reducing basis risk, and governing the data path from event to payout.
MLOps Systems Blueprint for Reliable AI
Production ML behaves like a three-body problem: code, data, and live behavior all pull in different directions. This guide shows how to turn that motion into a stable, self-correcting delivery loop.
Neural Architectures Decoded: FFNN, RNN, and Transformers
Feedforward nets, RNNs, and transformers are three different ways of teaching machines to notice pattern: layers for shape, recurrence for memory, and attention for selective focus. This guide compares them without losing the math.
DevOps to MLOps: Building the Shared Delivery Muscle
DevOps taught teams to ship code like a disciplined factory line; MLOps adds a third moving part, data, and suddenly the factory floor shifts under your feet. This guide shows what transfers cleanly and what breaks.
LLM Fine-Tuning: LoRA, QLoRA, DPO, and Mixture-of-Experts
A base LLM is a general instrument; fine-tuning changes how tightly it resonates with your task. This guide maps the adaptation spectrum from prompting to MoE, with the math behind each trade-off.
Federated Learning: Training Models Without Moving Data
Federated learning flips the usual gravity of ML: instead of hauling sensitive data to one warehouse, it sends the model out like a traveling teacher and brings back only the lessons. This guide explains the math and the operational trade-offs.
Data Warehouse, Data Lake, and Lakehouse: A Visual Architecture Guide
Warehouses, lakes, and lakehouses are really three answers to one question: when should raw data be forced into shape? This guide turns that architectural choice into concrete diagrams and decision rules.
Responsible AI: Safety, Fairness, and Trustworthy Systems
Getting a model to work is only the opening scene; the harder plot begins when it must stay fair, explainable, safe, and accountable under pressure. This guide maps the pillars and practices that keep trust from collapsing.
AI Governance and Regulations: From EU AI Act to ISO 42001
AI governance is the moment the story meets law: models leave the lab and enter a world of risk tiers, audits, and named obligations. This guide maps the major frameworks and what they require teams to actually build.
Retrieval-Augmented Generation: Architecture, Evaluation, and Production
RAG gives an LLM a memory it can check instead of bluffing from a frozen past. This guide follows the full pipeline from chunking to evaluation so a prototype can grow into a production system.
AI Agents: From ReAct to Multi-Agent Systems
An agent is what happens when an LLM stops answering once and starts acting repeatedly in the world. This guide traces the control loops, tool use, and guardrails that separate a demo agent from a dependable one.