Mistral 7B and NLLB in Docker: model packaging, inference boundaries and data-location trade-offs.
EXPLORATIONS
Engineering beyond client assignments.
Hands-on experiments around AI, automation, infrastructure and recoverability, clearly separated from commercial client work.
THE THINKER · AI & JUDGEMENTExperiment areas
MistralNLLBDocker
Explore ↗Wake word, Whisper/faster-whisper speech recognition, Gemma, OpenVoice/MeloTTS/Piper. Latency, permissions and local processing are central challenges.
SpeechLLMPrivacy
Explore ↗Evaluation of memory, availability, operational cost, data sensitivity and provider exit options.
SovereigntyGPUCloud
Explore ↗Cold backups, integrity checks, automated validation and functional restoration drills on a personal infrastructure. Not represented as a customer assignment.
ReliabilityRecoveryDocker
Explore ↗Prometheus, Grafana, Thanos, logging, service checks and actionable alerts for applications and infrastructure.
MetricsSLOMonitoring
Explore ↗SonarQube, Trivy, CI/CD controls, security awareness workshops and release governance.
DevSecOpsCI/CDSecurity
Explore ↗