At Datacrunch Global, I developed and maintained production features using React and Django, and diagnosed service outages. I also documented development and deployment procedures.
For Bo, my personal search engine, I built a hybrid retrieval pipeline for local files using BM25, BGE, RRF, and Cross-Encoder reranking. I created an evaluation set with Recall@k, MRR, and nDCG to identify failure modes and guide improvements.
I also built an evaluation framework for agent harnesses and inference configurations, comparing quality, latency, and cost. Earlier, as an Analyst at Eilers & Krejcik Gaming, I built automation tools for repetitive Excel-based research and introduced scripting workflows to non-technical teams.
