👨‍💻 Bernard Ladenthin | Homepage | LinkedIn

I'm a Senior L3 Site Reliability Engineer at Deutsche Bank in Berlin, focused on stabilizing mission-critical financial systems, cloud-native microservices, and production reliability at scale.

I spent over a decade at IAV GmbH as a Senior Software Developer, working on... automotive infotainment and embedded systems with a strong focus on:

• Android-based HMI development (AOSP)
• CI/CD pipelines and test automation
• Performance optimization and system integration
• Automotive standards (ASPICE, safety-critical systems)

• JVM: Java, Kotlin, Groovy
• High-Level: C++, .NET, PHP
• Scripting: Python
• Low-Level: Assembler (x86, Java bytecode), C
• OS & System: GNU/Linux, LAMP, WINE
• Databases & Data Integrity: MySQL, SQL, LMDB (key–value store), data integrity (ECC, FEC, RS-Codes)
• Performance: Optimization, high-performance computing (OpenCL, VHDL)
• Cryptography: ECC, RSA, Diffie–Hellman, blockchain tech (Bitcoin, Monero, Bitmessage)
• Testing: Low-level & high-level testing, test-driven architecture
• Build & DevOps: Maven, Ant, Gradle, Jenkins
• Distributed Computing: Parallel/distributed processing
• Privacy & Security: TOR, secure communications, blockchain privacy, attack vector analysis

• Automotive HMI software development
• Software integration and interface design
• Embedded systems development and protocol-level work
• DevOps for infotainment systems and continuous delivery
• Guest lecturer at universities on CI/CD in automotive context

• ⚡ Certified for working on high-voltage automotive systems (up to 1000 V AC / 1500 V DC):
  Elektrofachkraft (EFffT) gem. DGUV-I 209-093 – Stufe 2E (FHV)
• 🩹 Certified first aider
• 🎓 Guest lecturer on DevOps and continuous delivery in automotive software

A high-performance JVM + OpenCL tool that generates and checks Bitcoin & altcoin addresses — designed for cryptographic edge-case exploration and performance experiments.
Focuses on:

• Parallel EC key generation
• Vanity address search
• Efficient LMDB database integration

A test-driven, thread-safe streaming buffer that connects an OutputStream to an InputStream — enabling real-time, bidirectional data flow.
Key features:

• Supports concurrent reads and writes
• Automatic buffer trimming and memory optimization
• Optional safe-write mechanism (clone-on-write for immutable data)
• Suitable for embedded systems, protocol parsers, and streaming pipelines

Java bindings for llama.cpp — enabling local LLM inference (text completion, chat, embeddings, reranking, infilling) directly from the JVM without cloud dependencies.
Key features:

• Blocking and streaming text/chat completion with full sampling control
• OpenAI-compatible chat completion with tool/function calling support
• Embeddings, reranking, and fill-in-the-middle (infilling)
• Pre-built native binaries for Linux, macOS, Windows; CUDA, Metal, and Vulkan via local build
• Android support (CPU and OpenCL/Adreno classifiers)
• Published on Maven Central as net.ladenthin:llama

A Maven plugin that generates hierarchical, AI-readable documentation of Java source projects using local llama.cpp-compatible models — no cloud dependency required.
Key features:

• Per-file .ai.md summaries with keyword metadata extracted by a local GGUF model
• Package-level aggregation into package.ai.md for fast semantic navigation
• Hash-based incremental updates (skips unchanged files)
• Configurable prompts for summaries and keywords
• Published on Maven Central as net.ladenthin:llamacpp-ai-index-maven-plugin

I love working on electronics projects, exploring reverse engineering, and experimenting with low-level optimization techniques.
In my spare time, I craft utilities and microtools that blend software and hardware.

Based near Berlin, Germany, I'm passionate about well-tested, efficient, and secure software — whether it's running on a car, a GPU, or a tiny embedded chip.

Feel free to connect with me on LinkedIn.
I'm always happy to talk about tech, performance, or embedded systems!