Projects
A selection of things I've built — from agentic systems and RAG pipelines to low-level runtimes and developer tooling.
Multi-Agent Orchestrator
A LangGraph-based system that coordinates specialized agents to plan, research and execute long-running tasks with shared memory.
RAG Knowledge Base
Production retrieval-augmented generation pipeline with hybrid search, re-ranking and evaluation harnesses for grounded answers.
LLM Fine-Tuning Toolkit
Reproducible fine-tuning workflows with dataset curation, LoRA training and automated regression tests via pytest.
Prompt Evaluation Suite
A framework for systematic prompt engineering with versioning, A/B comparisons and structured scoring across models.
C++ Inference Runtime
A lightweight C++ runtime for serving small models with low latency and a clean Lua scripting layer for orchestration.
Agent Eval Dashboard
An internal dashboard to trace agent runs, inspect tool calls and surface failure modes across multi-step pipelines.