Section 1: The dataset and the problem
The hidden distribution, the dataset I actually get, and why generative modeling starts there.
Writing
Writing
Notes on AI platforms, agent quality, and how software engineering is changing.
Series
Math behind Generative AI
I am working through Mathematical Foundations of Generative AI and writing a deeper note for each section, with the visual and the detours that make the formula hold.
Section 2: The general principle of generative models
Pick a parametric family, define a divergence, and minimise it using only samples.
Section 3a: f-divergences
One convex function f turns a density ratio into a divergence. Forward KL, reverse KL, Jensen–Shannon, and total variation are four choices of f.
Section 3b: The convex conjugate and the variational bound
Rewrite D_f using only sample averages. The convex conjugate turns the density ratio into a critic game — the GAN objective in disguise.
Section 4a: From the variational bound to a trainable min-max game
Turn Week 1's abstract sup over functions T into a neural-network critic T_w, creating the two-player saddle-point game J(theta, w).
2026
October
Sixteen years in, AI writes my code. Here’s what I’m still for.
One engineer’s journey through faster code, rising AI bills, and the work that still needs a human.
What 3Blue1Brown’s case for math teaches software engineers in the age of AI
AI will not make software engineers obsolete. It moves the value of the work from writing code toward judgment, trust, and people.
Section 4a: From the variational bound to a trainable min-max game
Turn Week 1's abstract sup over functions T into a neural-network critic T_w, creating the two-player saddle-point game J(theta, w).
Section 3b: The convex conjugate and the variational bound
Rewrite D_f using only sample averages. The convex conjugate turns the density ratio into a critic game — the GAN objective in disguise.
Section 3a: f-divergences
One convex function f turns a density ratio into a divergence. Forward KL, reverse KL, Jensen–Shannon, and total variation are four choices of f.
Section 2: The general principle of generative models
Pick a parametric family, define a divergence, and minimise it using only samples.
Section 1: The dataset and the problem
The hidden distribution, the dataset I actually get, and why generative modeling starts there.
Beyond Text Streams: Building Interactive Micro-Frontends with MCP Apps
Why conversational AI needs rich visual components, and how to embed sandboxed interactive widgets into agent chat using MCP.
September
ISO 42001 for Agentic AI: An Architecture Guide to the AI Management Standard
What the ISO/IEC 42001 standard means when systems move from passive text generation to autonomous, tool-using agents.
Platform Engineer with IssueOps and GitOps
How platform engineering eliminates infrastructure bottlenecks while automating enterprise security, compliance, and cloud onboarding through issue-driven GitOps.
2026
August
EKS Auto Mode for Agentic AI Platforms: The Case For and Against
Why platform engineers building agentic AI systems choose Amazon EKS Auto Mode over Fargate and ECS, and the real-world trade-offs to watch out for.
July
Measure Once, Distribute Everywhere: A Central Telemetry Engine for GenAI
How to handle high-cardinality GenAI traces by ingesting once with OpenTelemetry and forking into APM, billing, and ClickHouse.
Continuous Compliance as Code: Automating SOC 2 and ISO 27001 in GitOps
How to eliminate audit spreadsheets by automating branch protections, code ownership validation, and continuous audit evidence collection.
June
The Token Vault: Secure OAuth Delegation and Identity for AI Agents
Why AI agents need their own first-class cryptographic identity alongside user delegation, and how a Token Vault enables secure tool execution.
2025
December
Resilient Multi-Cloud LLM Gateways: Routing, Rate Limits, and Failovers
How to design a production model gateway that handles provider outages, rate limits, and latency spikes across multiple cloud providers.
October
Network Segmentation for AI Systems: Containing Blast Radiuses in Agent Architectures
Why flat virtual networks are dangerous for AI systems, and how to design isolated network zones to prevent SSRF and data exfiltration.
August
Why Traditional API Gateways Break with Generative AI
How streaming tokens, minute-long connections, and token-based rate limits break standard enterprise API proxies.
Responsible AI for Autonomous Systems: Principles Beyond Traditional Chatbots
Why safety frameworks must shift from regulating text output to constraining real-world actions when agents use tools.
June
Building an Enterprise MCP Tool Broker: Turning OpenAPI into Agent Tools
How to connect AI agents to hundreds of existing internal APIs dynamically using the Model Context Protocol.
May
Solving for X, Part 2: Tackling Tougher Linear Equations in Data Science
Tackling underdetermined and overdetermined linear systems in data science with practical Python code, rank analysis, and least squares solutions.
2025
May
Solving for X: Cracking the Code of Linear Equations in Data Science
How to formulate and solve systems of linear equations in data science using matrix representations and the fundamental AX = B formulation.
Getting a Grip on Linear Algebra for Data Science: It's Not as Scary as It Sounds!
A practical guide to linear algebra fundamentals for data science, covering matrices, data organization, and finding linear relationships.
April
Making Sure AI Agents Play Nice: A Look at How We Evaluate Them
An in-depth analysis of evaluation frameworks for AI agents, covering conversational, autonomous, and multi-agent systems with practical insights and methodologies.
The Future of Coding is Getting Wild: How AI, Like This Cursor IDE Thing, is Changing Everything
An exploration of how AI is transforming coding, with a focus on Cursor IDE and the changing landscape of software development.
March
Building Scalable Agentic AI Platforms: A Technical Deep Dive - Part 2
A technical deep dive into the four core platform modules: LLM gateways, MCP tool brokers, secure execution sandboxes, and unified telemetry.
Building Scalable Agentic AI Platforms: A Technical Deep Dive - Part 1
An in-depth exploration of architectural patterns and best practices for building enterprise-grade AI platforms that scale.