Hi, I'm Will. I recently graduated from Johns Hopkins University, where I moved from a Bachelor's in Chemical Engineering to a Master's in Computer Science.
I like exploring what frontier AI can do, how to channel it into specific domains, and how to keep it reliable as it gets more capable. I like building the systems that bring it to life: agents, data pipelines, and evaluation infrastructure.
Experience
- >Machine Learning ResearcherDr. Nicholas Andrews · Human Language Technology Center of Excellence, Johns Hopkins
Training deep research forecasting agents for general event prediction. Assembled a 95M-document, time-versioned Wikipedia index and wrote the async scraper that turns real-world data into forecasting questions across dozens of sources, from sports and finance to clinical trials and weather.
Distributed Data EngineeringLLM EvaluationFall 2025-Presentlab - >Software Development Engineer InternAmazon
Built an LLM inference pipeline on Amazon Bedrock that generates product-catalog content across hundreds of thousands of products, and shipped it to users.
Production MLAWS InfrastructureSummer 2025 - >Biopharma ResearcherBetenbaugh Lab · Advanced Mammalian Biomanufacturing Innovation Center, Johns Hopkins
Genetically engineered CHO cell lines, the cells the industry uses to manufacture antibody drugs, to increase monoclonal antibody (mAb) yields. Also worked on fungal bioleaching, recovering metals from waste.
BiomanufacturingWet Lab Research2022-2024lab
Projects
- >Agentc
Cuts the token cost of multi-step agent pipelines without changing application code. Traces every LLM call, reuses the repeated ones, and rewrites the expensive paths on its own.
Cost OptimizationRust + PythonOngoinggit - >Fly Orchestration Kit
Runs a team of AI coding agents in parallel: it coordinates handoffs between them, retries and escalates when steps stall, and ships a library of 50+ prompts and 118 skills that work across Claude, Codex, and Gemini. Built on Emanuel's Flywheel.
Prompt & Skill LibraryAgent CoordinationTwo-week project - >Entropy Wiki
Collects coding agent environments, orchestration frameworks, and reusable AI skills and prompts into one reference hub.
Agent EcosystemsSkill LibraryThree-day projectgit - >Mycelium
Organizes research and development notes in a personal knowledge vault and Obsidian-style workspace.
Note SystemsResearch LogsOne-week projectgit - >Research Intelligence Engine
Returns cited answers across 100K+ arXiv papers through a two-stage RAG pipeline: semantic vector search followed by cross-encoder reranking.
Scholarly SearchEvidence SynthesisTwo-week projectgit