Currently:
- 🧪 Researching compute-efficient model architecture at OpenAI's Training team
- 🔬 Understanding how vision-language models build syntactic representations
- ⤴ Scaling up neural satisfiability solvers
- 📈 Exploring how scaling laws scale with data complexity
- 👾 Fine-tuning LLM agents to play games with online RL
- ↩️ Replacing backprop in autoregressive language models
Previously:
- 🙈 Engineer #1 @ Reworkd working on multimodal code-generation for web data extraction
- 🎓 Graduated from Carnegie Mellon '23 with Honors in Computer Science
- 🔍 My thesis on vision-language semantics is cited by Google Brain, Meta AI, Stanford, etc.
- 📄 Published papers at ACL, ICLR, EMNLP, & EACL conferences and NeurIPS & ICCV workshops
- 🧑💻 Exited a content research micro-SaaS with some cool clustering, fact checking, & generation features
- 🤖 Fine-tuned language models at Microsoft AI over summer '22
- 🛠️ Worked on information retrieval, question answering, & summarization at various startups '20-21
- 🧠 Developed brain-computer interfaces with NSF funding and placed 1st nationally at NeuroTechX '20
- 🏆 Won 10+ hackathons including 1st @ Facebook '19, 2nd @ UCLA '19, 3rd @ MIT '20
Warning: has not learnt the Bitter Lesson. Prone to getting nerd-sniped by linguistically & cognitively motivated AI research directions.