to presentFull-time
Junior AI Developer
Fuseable Technologies, Bonifacio Global City, Taguig
- Build agentic AI features for a wealth management platform serving private banking clients, using Python, LangChain, and LangGraph.
- Designed an MCP-based architecture that exposes about 86 tools to LLM agents, giving them structured access to platform data and actions.
- Built a PII detection pipeline with Microsoft Presidio, spaCy, and Qwen3Guard, including support for Asian languages, to protect sensitive client data in LLM workflows.
- Integrated AWS Bedrock Guardrails with LangChain agents and set up model routing across AWS Bedrock and OpenRouter to balance cost, quota limits, and reliability.
- Implemented multi-level and semantic caching for LangChain agents to cut repeat LLM calls, latency, and cost.
- Engineered a portfolio drift detection system that flags allocation deviations for human-in-the-loop review, and extended it with ML-based scoring using XGBoost and LightGBM.
- Re-architected AI workflows from n8n into code-based Python and LangChain services, moving agent logic into version-controlled, maintainable code.
- Deploy and monitor AI services with Docker, AWS EC2, and Amazon ECR, and integrate agents into production frontend and backend applications.