Systems
AegisOS: what an AI system can promise when the network disappears
Local evidence, signed timelines, delayed transport, and the difference between a working offline prototype and a validated field system.
Archive · · 9 min
Writing / An open engineering notebook
Notes from
the work.
What I am building, what I am learning, and the questions that deserve a closer look. Notes on data systems, machine learning, and the decisions behind reliable software.
A record of the process: experiments, architecture, lessons, and ideas worth returning to.
Follow the learning journal- 47
- published articles
- 5
- topics
- 249
- minutes of reading
- Systems
- AI Systems
- AI & ML
- Backend
- DevOps
0
Published Articles
0
Categories
0+
Min Total Read Time
0
Months Writing
Since June 2024
Writing / In the spotlight
Research & engineering highlights.
15 selected essays on the decisions behind data systems, machine learning, and applied AI—from the first data contract to the evidence behind an answer.
AI Systems
Structura: preserving evidence before documents become embeddings
A source-to-answer design study of document identity, extraction, retrieval, and citations—and the information an embedding cannot preserve.
Archive · · 9 min
AI Systems
When retrieval should retry, abstain, or ask for review
A corrective retrieval protocol that treats missing evidence, contradiction, permissions, and exhausted budgets as different decisions.
Archive · · 9 min
AI Systems
Semantic caching when permissions and documents keep changing
Why similar questions are not necessarily equivalent requests, and how evidence lineage changes cache lookup, invalidation, and evaluation.
Archive · · 9 min
AI Systems
Governed Bedrock applications: tracing a request from identity to evidence
A request-level design for retrieval permissions, model access, tool authorization, network boundaries, and audit evidence in AWS.
Archive · · 10 min
Systems
Python, Rust, and Go at the boundaries of an agent system
A design study of deadlines, backpressure, tool contracts, and ambiguous failures before deciding whether a language boundary is worth adding.
Archive · · 9 min
AI & ML
When does a neural representation earn its complexity?
A study protocol comparing simple baselines, frozen embeddings, and learned representations under the same data, shift, and deployment constraints.
Archive · · 10 min
AI & ML
A better fraud score can hide a worse model
An experiment protocol connecting temporal splits, SMOTE, probability calibration, and the workload an investigator can actually handle.
Archive · · 9 min
AI & ML
One ML pipeline, three execution layers: PySpark, Polars, and Rust
A reproducible experiment for choosing distributed preparation, local iteration, and native processing around a SageMaker training workflow.
Archive · · 9 min
Backend
From JDBC to AI-ready data: migrating without breaking existing consumers
A migration design built around compatibility contracts, change capture, reconciliation, and a cutover that can actually be reversed.
Archive · · 9 min
Backend
The database boundary in an AI system
Follow one entity through Postgres, Redshift, Cassandra, and a vector index to make ownership, freshness, and semantic meaning explicit.
Archive · · 10 min
AI Systems
From biomedical text to research evidence
A methods note on biomedical concept extraction, ontology, provenance, and why matching a medical term is not the same as validating a claim.
Archive · · 9 min
AI Systems
Evaluating change impact when the dependency graph is incomplete
A general evaluation protocol for separating dependency coverage, impact ranking, and verification, with explicit limits on what each can establish.
Archive · · 9 min
AI Systems
Turning engineering questions into reproducible studies
A research agenda built around bounded claims, meaningful baselines, traceable experiments, and results that remain useful when a hypothesis fails.
Archive · · 10 min
AI & ML
Why Cryptographic Inventory Is Becoming a Core Layer for Data, LLMs, and AI/ML Pipelines
AI runs on data, but trust is what makes that data usable. As LLMs and AI pipelines scale, cryptographic inventory is becoming essential for visibility, governance, and long-term resilience.
· 2 min
What I write about
Topics & themes.
Field notes from building production AI and data systems — the ideas, trade-offs, and hard-won details behind the work.
Retrieval & RAG
Evidence-grade retrieval, corrective RAG, citation gates, and why atoms beat chunks.
RAGretrievalevidence
LLM Engineering & Agents
Multi-agent pipelines over MCP, model routing, grounding, and shipping agents that don't freelance.
LLMsagentsMCP
Data Engineering
Pipelines, warehouses, and moving real weight — Spark, Kafka, pgvector, and provenance at scale.
pipelineswarehousespgvector
Systems & Rust
Rust engines, distributed systems, determinism, and building software that holds up under load.
Rustsystemsperformance
Security & Trust
Cryptographic inventory, tamper-evident audit, zero-trust IAM, and provenance you can replay.
cryptozero-trustaudit
Cloud & Defense Tech
AWS GovCloud architecture, FedRAMP posture, and engineering for mission-critical environments.
GovCloudFedRAMPdefense
All articles
The archive.
Featured essays first, followed by the chronological archive. Dates shown are each article’s archive dates.
Read by research series47 articles · page 1/6
Newsletter
New posts, straight to your inbox.
Occasional, high-signal notes on retrieval, LLM systems, and building AI you can trust. No spam — unsubscribe anytime.
Join the list · one click to leave