Recent
Speculative Decoding and the Model Choice: Lessons
Speculative decoding model differences.
Inference-Aware AI AI EngineeringEngineering Best Practices
Standing Up vLLM on a Single A10G: From First Boot to Dual-Model Deployment
Deploying vLLM with docker on AWS using terraform.
AI EngineeringInference-Aware AI
More Articles
How to Lead High-Confidence, High-Certainty People Without Crushing Their Value
This article shows how leaders can channel strong, dominating confidence into team-strengthening collaboration.
Leadership
Rough Notes - Product Observability
A part of Product strategy that isn’t just about shipping features fast.
LeadershipEngineering Best Practices Software Philosophy
Inference-Aware AI: Working Definitions
A glossary of terms that define the concept of inference-aware agents, breaking down the core ideas, agent types, awareness dimensions, and platform components behind cost-efficient AI systems.
AI EngineeringInference-Aware AI Software Philosophy
A Hypothesis: Inference-Aware Agents Could Be the Next Big Leap in AI Efficiency
An introduction to the hypothesis that AI agents can be made faster, cheaper, and more effective through an inference-aware platform that optimizes how they decide, act, and use resources.
AI EngineeringInference-Aware AI Software Philosophy
A VECTR-Guided Refactor with Cursor.
Let's review a real example i've had to refactor to make it easier to contend with.
Software PhilosophyEngineering Best Practices Coding Practices
The Duplication Dilemma: A VECTR Guide to Repeating Yourself
This article clarifies the "Don't Repeat Yourself" (DRY) principle within the context of VECTR
Software PhilosophyEngineering Best Practices Coding Practices
Scaling Engineering with AI from 0 to 50
What it really takes to scale an engineering team from 0 to 50 inside a 100+ person company in today’s AI-native world.
AI EngineeringEngineering Team ScalingLeadership
Logistic Regression from Scratch with Python (Full Implementation)
Logistic regression from scratch with notes and learnings.
ML & Data Science
VECTR: Velocity-Engineered Code for Rapid Teams
VECTR is a pragmatic software philosophy focused on speed, clarity, and context. It favors useful code, justified abstractions, and adaptive design over rigid rules.
Software PhilosophyEngineering Best Practices
Notes to Self: Hiring Playbook
No-fluff hiring guide for engineering leaders
Engineering Team ScalingHiring & Talent
Multiple Linear Regression from Scratch (with Diagnostics)
A from-scratch implementation of multiple linear regression using gradient descent, with full diagnostic plots and batch prediction on test data.
ML & Data Science
Simple Linear Regression on Housing Data (Notes)
Just some note on linear regression to come back later to.
ML & Data Science