I co-created De(ep)Composition, a system for long-horizon reasoning that decomposes unstructured information into hierarchical components and stores it in persistent, queryable memory. I'm optimistic about moving beyond gathering large volumes of static information toward reasoning frameworks useful in settings that require continuously updated views grounded in accumulated evidence (e.g., hedge-fund portfolio management for long–short equity positioning based on earnings revisions, corporate teams monitoring markets and competitors, and investigative journalists maintaining long-running storylines across many documents over time).
I founded Airstitch, a subscription-based software product that operates inside iMessage (and WhatsApp), where much of everyday coordination already occurs. Airstitch connects to platforms like Email, Calendar, Notion, Slack, Google Sheets, Linear, Canvas, LinkedIn, and can go beyond an advisory interface typical of today's AI interfaces and can take action on the user's behalf.
I study Computer Science at Stanford. Some of my favorite classes have focused on the tools used to build decision-support systems for domains like air traffic control, robotic planetary exploration, and investing. I liked working on modeling portfolio rebalancing as a POMDP.
Another work I'm proud of is building ScalableTokenizer, a tokenizer that learns a vocabulary with column generation and decodes text with dynamic programming while consulting morphology-aware features, named-entity gazetters, and regex guards. Inspired by foundational tokenizers such as SentencePiece and Byte-Pair Encoding, I see work like this as a path toward more efficient representations of text in LLMs.
Previously, I worked at Tesla as an embedded systems software engineer and Foundry (now called Mithril), an AI compute infrastructure startup backed by Sequoia, Lightspeed, Redpoint, and NVIDIA Ventures.
I was an early hire (investment team member #3) of Phoenix Rising, a G2VP and Kleiner Perkins spinoff tech venture fund, where I focused on spatial intelligence and datacenters. I also worked on ecosystem building (co-hosted AI poker nights with Kleiner Perkins and built a new incubation program). Our portfolio includes 1X, Seegrid, Scandit, and Robinhood (pre-IPO).
More of what I've done previously can be found here.