Explored. Tested. Shared.
On AI systems design, data architecture, and the software frontier — insights from the engineers working at the edge.
Model Abliteration
Model abliteration: the process by which a model’s safety guardrails get bypassed, exposing knowledge it was designed to refuse. A walk through how easy it actually is, why prompt-engineering alone isn’t enough, and what threats expand as model access democratizes.
read article →Recent articles.
C++ in 2026: A Language That Talks Modern, Ships Legacy
or: how "import std;" — chapter 1 of a good book — turned into a rabbit-hole of building LLVM from source. C++'s specification has kept pace with modern ideas, but it's bolted onto a tooling ecosystem that hasn't — and beginners pay for that gap with a nightmarish developer experience
read article →Rethinking Process Modeling in the GenAI era
GenAI has done something quiet but decisive: it's made coding cheap. That single shift pulls the foundation out from under low-code, no-code, and the citizen developer persona built to justify them — along with the BPMN, DMN, CMMN, and XPDL standards that were never quite as portable or maintainable as promised. What replaces them isn't a better canvas, but code-first, spec-driven development backed by durable, scalable runtimes and real observability. Platforms like Temporal are already showing what this looks like in production.
𝗣𝗵𝘆𝘀𝗶𝗰𝘀 𝗺𝗲𝗲𝘁𝘀 𝗘𝗹𝗲𝗰𝘁𝗶𝗼𝗻𝘀!!
Physicists just proved "trustless" quantum voting works on a lab bench with entangled photons — a real conceptual milestone, but scaling from 14 photons to millions, unresolved security trade-offs, and unaddressed classical alternatives mean it's decades (if ever) from replacing a real ballot box.
From the archive.
Encountering eigenvalues in college vs. now using LLMs as patient explainers. On the joy of learning.
Sarvam-M vs. LLaMA-3 vs. Mistral on Hindi, Tamil, Telugu. Measured latency, cost, token distribution.
An intern’s walkthrough of using LLMs to automate contract data extraction in transport & logistics — turning unstructured PDFs into structured digitized data, and what it took to get there.
Cut LLM spend ~5× with vector-based caching and routing.
Dijkstra on natural-language programming, applied to vibe coding.
Building a healthcare RAG pipeline from zero, honestly.
Alternatives to matrix multiplication for low-power edge GenAI.
A custom Metal device plugin for GPU slicing on k3s/k8s.
Natural-language SQL queries with GPT-3.5 against CockroachDB.
Working on something we should write about?
Tell us what you’re building. The best perspectives come from real engagements.