Technical articles and historical engineering writing on software and systems engineering, real-time data processing, artificial intelligence, speech and signal processing, digital forensics, cybersecurity and algorithms.
The Blog contains my technical articles on software and systems engineering, artificial intelligence, digital forensics, cybersecurity, and the history of technology.
Why Java virtual-thread concurrency and concurrent JDBC capacity are different quantities, with HikariCP, backpressure, and queueing latency as the operational boundary.
How voice activity detection, minimum speech duration, silence gaps, and segment caps jointly shape endpoint latency and inference efficiency in real-time ASR.
A technical provenance model that separates original evidence, deterministic transformations, model inference, and analyst interpretation in digital-forensic workflows.
A technical guide separating deasciification, character normalization, Numeric Soundex, edit distance, and syllabification by the problem each method actually solves.
A historical study of the origins of computer, algorithm, byte, gigabyte, robot, mechatronics, cybernetics, bug, Web, Bluetooth, Python and artificial intelligence, tracing engineering metaphors, standards and computing culture through verifiable sources.
A systems-engineering treatment of sustainable ASR capacity through RTF, segmentation, VAD, queueing, backpressure, CPU/GPU sharing and tail latency.
My IBM POWER9 AC922 has been a long-running engineering platform since 2019 for digital forensics and AI work involving AltiVec/VSX, OpenMP, CUDA, Tesla V100, dlib, MXNet, FAISS, ArcFace, Kaldi, Vosk, whisper.cpp, OCR, file carving, and current CTranslate2/faster-whisper experiments.
Mean latency is not enough for Java performance work; queues, allocation rate, garbage collection, lock contention, and downstream delays shape P99 and higher-percentile behavior.
An ArcFace-trained ONNX face-embedding model cannot be moved directly to CTranslate2 merely because its graph is available in ONNX; the supported model architecture and inference graph must also match.
Atomic rename, visibility, and durability are separate guarantees in file publication; reliable workflows must account for temporary files, fsync, directory persistence, and NFS visibility.
HikariCP pool size should follow sustainable database concurrency rather than application request concurrency; a larger pool can increase contention without increasing throughput.
Retry is safe only when transient-failure classification, idempotency, total time budget, and queue limits are defined together.
Uses only this site's published content. How it works
