Prague, Czechia

Thanh Dat T.

Java / Spring backend engineer — applied AI: RAG · agents · LLM integration

LinkedIn Open to AI engineering roles

Java/Spring backend engineer with 4 years of production experience across enterprise banking (Komerční banka) and custom software delivery (Syntea). Strong in microservices, messaging, and production reliability in regulated environments. Over the past year I have been building applied AI systems hands-on: a retrieval pipeline over ~600k Czech court decisions, multi-agent automation with hosted and local models, and self-hosted GPU inference. Looking for an AI engineering role in Prague where solid backend depth meets applied AI.

Experience

Java Software Engineer · Syntea software group a.s.

Prague · 2025 – 09/2026

  • Migrated legacy internal libraries from Java 8 to Java 21.
  • Designed and implemented a monitoring solution for internal services.
  • Led the redesign and implementation of the internal services gateway.
  • Optimized read-heavy REST endpoints, cutting response times from tens of seconds to near-instant.
  • Implemented POST endpoints for automated worklist task creation, resolving race conditions and server-side errors.
  • Delivered new API functionality on a Spring Boot / Oracle stack, verifying contract correctness against technical specs.

Software Engineer (Java Developer) · Komerční banka

Prague · 2022 – 2025

  • Contributed to the maintenance and bug fixes of KB's internal development platform (SPEED). Designed, developed, and maintained multiple in-house microservices.
  • Provided production support to ensure system stability and performance.

Selected projects

Czech case-law RAG

Retrieval-augmented search over Czech Ministry of Justice open data (~600k court decisions): ingestion, embeddings, vector search (Qdrant), and citation-checked answer evaluation.

Multi-agent automation lab

Self-hosted agents that run real work: multi-provider LLM routing (hosted APIs + local GPU models), usage and cost tracking, scheduled pipelines, knowledge-base tooling with validation gates.

POS platform modernization (EET 2.0)

Modernizing a legacy point-of-sale application for the Czech EET 2.0 regime (touch UI, Linux & Windows builds) and its operations backbone: HTTPS provisioning, WireGuard fleet network, versioned auto-update feed — rolling out across shops.

Private cloud & observability

Self-hosted file/office cloud, centralized logs and metrics (Grafana, Loki, Beszel), reverse proxy with automated TLS, WireGuard networking — home hardware with a VPS edge.

Local-GPU speech pipeline

Text-to-speech dubbing for serialized audio fiction (Qwen3-TTS on Intel Arc): voice design, scene-aware splitting, batch rendering.

Skills

Backend
Java 8–21, Spring Boot, REST APIs, Apache Kafka, ActiveMQ Artemis, Camunda BPM, PostgreSQL, Oracle, SQL, Elasticsearch, JUnit, WireMock, feature flags
Platform & tooling
Kubernetes, Docker, Jenkins, Git / Bitbucket, monitoring & observability, CyberArk
AI / LLM (hands-on)
RAG pipelines (embeddings, vector DBs, retrieval evaluation), LLM integration & provider abstraction, agent orchestration, local inference (llama.cpp, vLLM, Intel Arc GPU), TTS pipelines
Systems
Linux, nginx / Caddy, WireGuard, Tailscale, TLS / ACME automation
Languages
Czech · English · German — professional working proficiency

Education

Unicorn University, Prague — IT · 2018 – 2023 (studies interrupted for a full-time role at Komerční banka)

University of Finance and Administration, Prague — Bachelor's degree, Banking · 2015 – 2018