# Janaka Premathilaka

> Senior Java Engineer & Solution Architect, Zug, Switzerland. Banking-grade Java engineering, now bringing practical AI into regulated systems.

Every page linked here is plain HTML that reads without JavaScript. The same facts are available as JSON under /api/public/v1/ and as Schema.org JSON-LD in each page.

## Profile

- [Home](https://janaka.me/): 22+ years of banking-grade Java at UBS, Credit Suisse and the European Commission. Now bringing LLM, RAG and agentic AI into regulated systems. Zug.
- [CV](https://janaka.me/resume/): experience, certifications, education and stack, English and German
- [Profile JSON](https://janaka.me/api/public/v1/profile.json): the CV as structured data
- [Contact](https://janaka.me/#contact): email janaka2@gmail.com

## Projects and products

- [AssetCare](https://janaka.me/lab/assetcare/): A production-deployed Java 25, Spring Boot, Angular and Kubernetes application demonstrating modern full-stack, cloud-native and AI-assisted software engineering.
- [BlitzMath](https://janaka.me/products/#blitzmath): Live maths battles for classrooms and families: a teacher or parent hosts, children join with a six-digit code on their own device, and everyone answers the same questions at the same time. No accounts, no ads, no chat.
- [nüchtern](https://janaka.me/products/#nuechtern): Fasting companion that keeps the biology visible: phase timer, hydration, electrolytes. Offline, German and English.
- [Daily Momentum](https://janaka.me/products/#daily-momentum): A quiet record of where the day went: one tap says what you are doing, everything else is derived. On-device only.
- [Loop](https://janaka.me/products/#loop): Remembers what repeats and learns the rhythm from your own history, on your device or your own Drive.
- [BabyLoop](https://janaka.me/products/#babyloop): Two-tap care log for babies and toddlers, with a daily summary and a PDF for the paediatrician. On your device or your own Drive.
- [LeftHere](https://janaka.me/products/#lefthere): Take a photo and it remembers where you left the car, keys, bag or a favourite place; the map walks you back. On your device or your own Drive.
- [MP3 Alarm Clock](https://janaka.me/lab/#mp3-alarm-clock): A desktop alarm clock that plays an MP3 or your own recorded voice at a scheduled time, keeps the computer awake and wakes it from sleep, with family schedules and a phone remote. One Python file, macOS, Windows and Linux.

## Blog

- [Blog](https://janaka.me/blog/): Clear, hands-on articles on Java, Kafka, RAG, agents and evals.
- [BlitzMath: A Live Classroom Maths Game with One Durable Object per Room](https://janaka.me/blog/posts/blitzmath-realtime-classroom-game.html): How BlitzMath runs live maths games for a whole class on Cloudflare Workers: one Durable Object per room as referee, alarms as the clock, no database or accounts.
- [From Idea to Production: Building AssetCare with Java 25, Spring Boot, Angular, Kubernetes and AI](https://janaka.me/blog/posts/assetcare-idea-to-production.html): How a Java 25, Spring Boot 4.1 and Angular 22 application went from a prototype to a live Kubernetes deployment on Hetzner with Helm, Keycloak, PostgreSQL and Let's Encrypt, what broke on the way…
- [Dead-letter topics in Spring Kafka: retries and replay](https://janaka.me/blog/posts/spring-kafka-deadletter.html): How to handle poison pills in Spring Kafka with DefaultErrorHandler, @RetryableTopic, DeadLetterPublishingRecoverer and a governed replay path.
- [RAG with FAISS: Chunking, MMR and Re-Ranking](https://janaka.me/blog/posts/rag-faiss-patterns.html): Why naive top-k retrieval fails, how to chunk and index with FAISS, and how MMR and cross-encoder re-ranking fix it for a regulated bank.
- [Machine Learning: A Comprehensive Top-Down Guide](https://janaka.me/blog/posts/machine-learning-blog.html): A top-down introduction to machine learning: the workflow, the algorithms and the evaluation.
- [Evals — Designing and Interpreting AI Evaluation Frameworks](https://janaka.me/blog/posts/ai-evaluation-frameworks.html): Building and reading AI evaluation pipelines: metrics, tooling and agentic behaviour.
- [MCP for Agent Integrations: the Model Context Protocol](https://janaka.me/blog/posts/mcp-agent-integration.html): What the Model Context Protocol is, its three primitives and two transports, a minimal Python server, and how to run it safely in a regulated bank.

## AI

- [AI](https://janaka.me/ai/): Agents that are readable, reliable and worth shipping.

## Lab

- [Lab](https://janaka.me/lab/): Experiments and demos, runnable, with notes.
- [AssetCare case study: Java 25, Spring Boot, Angular and Kubernetes in production](https://janaka.me/lab/assetcare/): AssetCare is a production-deployed Java 25, Spring Boot, Angular and Kubernetes application demonstrating modern full-stack, cloud-native and AI-assisted software engineering. Architecture…
- [LLM Study Notes — Interactive Highlighter Toolbox](https://janaka.me/lab/Notes/llm-study-notes-handwritten-style.html): Interactive study notes on assessing large language models with a highlighter toolbox, sticky notes, and explain, analogy and ELI5 helpers.
- [LLM Notes — Hand‑Annotated Highlights Demo + Toolbox](https://janaka.me/lab/Notes/llm-hand%E2%80%91annotated-demo5.html): Interactive hand-annotated notes on large language models with a highlighter toolbox: mark up, add sticky notes, save, load and export your annotations.
- [LLM Study Notes — Assessing LLMs, Handwritten Style](https://janaka.me/lab/Notes/hand-book-note3.html): Handwritten-style study notes on assessing large language models: why they are large, how self-attention works, and why hallucination is the big problem.
- [Study‑Style Notes: Aspects & Assessment of LLMs](https://janaka.me/lab/Notes/hand-book-note2.html): Study-style notes on assessing large language models: factuality, reasoning, safety, latency and cost, with the retrieval, QA and generation metrics.
- [LLM Notes — Hand‑Annotated Highlights Demo](https://janaka.me/lab/Notes/hand-book-note1.html): Hand-annotated study notes on what large language models are used for: retrieval and RAG, content generation, reasoning, coding and agents.

## Academy

- [Academy](https://janaka.me/academy/): Engineers who surpass their teacher: daily notes and learning paths.
- [From code to production: building AssetCare — a learning series](https://janaka.me/academy/assetcare/): A fourteen-module learning series on the AssetCare reference application: architecture, Spring Boot, Angular, PostgreSQL and Liquibase, Keycloak, Docker, Kubernetes, Helm, Terraform, HTTPS, CI/CD…
- [From CRUD to production: Angular + Spring Boot, served as a seven-course meal](https://janaka.me/academy/production-ready-spring-angular/): AssetCare, a complete Angular 22 + Spring Boot 4.1 reference application: architecture, PostgreSQL migrations, a production API, the SPA, tests and scans, observability, Helm, K3s on one small…
- [Prompt Engineering Handbook](https://janaka.me/academy/modules/2026/FSE/promting/): Prompt engineering handbook for software engineers: foundations, templates and strategies for LLMs, agent workflows, RAG, debugging and daily coding.
- [Practical Quantum Entanglement and Quantum Computing](https://janaka.me/academy/modules/2026/FSE/entanglement/): Quantum entanglement and quantum computing explained simply for software engineers: what entanglement means, real use cases, current limits and a career roadmap.
- [Linux CLI Mastery](https://janaka.me/academy/modules/2026/FSE/Linux/): Linux CLI handbook for backend engineers: top commands, debugging commands, awk and sed, file handling, log investigation, one-liners, vi and troubleshooting.

## Optional

- [Full version of this file](https://janaka.me/llms-full.txt): profile, experience and every page summary
- [Public JSON catalogue](https://janaka.me/api/public/v1/index.json)
- [Atom feed](https://janaka.me/feed.xml)
- [Sitemap](https://janaka.me/sitemap.xml)