AI Developer
Questronix Corporation
Makati City
February 2026 – Present
- Built and maintain on-prem RAG chatbots on Dify using Qwen LLMs with Qwen and Nomic embeddings, serving 198 internal users across 423 conversations and 2,100+ messages in 3 months.
- Improved retrieval quality through system prompt tuning and retrieval configuration, increasing successfully answered queries by 90% and sharply reducing "insufficient knowledge" responses.
- Manage LLM serving on a GPUStack cluster, leading log analysis and incident investigation; diagnosed OOM-killed worker containers and recommended memory upgrades to the hardware team.
- Maintain and extend a Document Pre-Processor that converts 10 file formats (PDF, DOCX, PPTX, XLS/XLSX, CSV, HTML, XML, MD, TXT) into Markdown chunks for embedding in Dify.
- Sole deployer of Dockerized applications to production, handling all releases during bug fix and feature cycles.
- Replaced hardcoded API filtering in a Django backend with dynamic filtering configurable through Django Admin, and added backend aliasing and automatic chart coloring to remove manual frontend setup.
- Resolved 11 critical production bugs and shipped 2 features in 2 weeks, accounting for 90% of critical fixes during the support phase.
- Lead a 4-person software team, owning architecture decisions, task allocation, and mentoring of junior developers.
- Develop and maintain production Django applications.
- Develop and maintain production Python scripts.
Technology Exposure
AI Chatbots
RAG
LLMs
Dify
Qwen
Nomic Embed
Python
Django
SQL
Rancher
Kubernetes
Longhorn
ClickHouse
GPUStack
Docker
Unix