AI Service Full-stack Developer · PL
I connect the entire product flow—from GenAI, RAG, and STT to Flutter apps, web interfaces, Python/FastAPI backends, and production automation.
Selected work
I design beyond model integration, covering the real user journey, failure recovery, data, and deployment. The work below is selected from projects supported by source code and project documentation.
Built a web service that uploads documents containing text, images, tables, and equations, tracks indexing status, and answers questions grounded in selected files. The actual RAG engine and a demo mode are separated for flexible development and validation.
Users photograph multiple price tags and an LLM Vision API extracts product names, prices, quantities, and unit prices to rank comparable items. I implemented image processing, response parsing, local recovery, private usage logs, Android 15 support, and release delivery.
View on Google Play ↗
More experience
Filter by discipline to review relevant work. Each item summarizes only implementation verified through source code, documentation, or Git history.
Worked across requirements traceability and WBS management, mobile meeting notes, AI chat, an STT-to-LLM pipeline, agents, and mobile security integrations.
Implemented an asynchronous WhisperX service with large-audio segmentation, parallel processing, ETA calculation, result caching, and restart recovery.
Connected relevant-file selection, change planning, approval-based patching, validation, and failure analysis with a VS Code extension and offline VSIX distribution.
An operations-focused system using external REST/WebSocket APIs with staged execution, failure rollback, restart recovery, live charts, and alerts.
Extended an open-source research project with BLE RSSI stabilization, Geofencing and sensor-based power saving, a watchdog, and scan failure recovery.
Turned a Figma-based time planning interface into a React service with Supabase, drag and drop, SEO prerendering, and deployment automation.
Capabilities
I consider the operating conditions from the initial architecture so the interface, API, model, data, and deployment do not become disconnected pieces.
Designing document lifecycles, multi-turn queries, STT processing, and agent tool calls as complete user journeys.
LLM · RAG · STT · WhisperX · Vision · EmbeddingBuilding complex experiences—recording, files, chat, and status tracking—across Flutter apps and modern web interfaces.
Flutter · Dart · React · Next.js · Vue · NuxtCreating REST/WebSocket APIs and data structures for asynchronous work, long-running jobs, external APIs, and live connections.
Python · FastAPI · PostgreSQL · Supabase · RedisTreating progress, retries, caching, recovery, and logs as product features, supported by tests and deployment documentation.
Docker · Playwright · pytest · systemd · GitHow I work
From initial alignment to validation and handover, I keep decisions and evidence visible throughout the project.
Separate functional and non-functional needs and define done.
Identify external integrations, data, performance, and scope gaps.
Connect interface, API, data, and operations in one design.
Share working increments early and record changes and decisions.
Deliver core tests together with deployment and operations guides.
In an initial discussion, I will focus on the key user journey, existing systems and integrations, data and security constraints, timeline, and acceptance criteria.
Back to the top ↑