Backend / infrastructure software engineer in San Francisco. Around 2+ years building production systems (Qualtrics, Callingsai), plus an MS in AI/ML. Focused on backend and ML-platform / ML-infra work, the bridge between systems and machine learning.
🎓 MEng, Artificial Intelligence & Machine Learning, George Washington University (2026). BS Computer Science, Columbia University.
- Backend systems at scale. A tier-1 access-control service at 5M+ requests/day, a FedRAMP-compliant async log pipeline, and CI/CD tooling at Qualtrics.
- Leading and shipping. Led 4 engineers under the CTO at Callingsai; shipped 3 production systems in under 8 months.
- Production ML. A Two-Tower PyTorch matching microservice (sub-100ms queries via PostgreSQL pgvector, full CI/CD to AWS), and RAG agents grounded in SQL / Neo4j knowledge graphs.
LINet, Multi-Stream Networks · 📄 arXiv:2606.31135: a novel multi-stream multi-modal fusion architecture, designed, built, and benchmarked from scratch in PyTorch. Custom LIConv2d operator integrating modalities at every convolutional layer. Strongest mean class accuracy of any model trained from scratch on SUN RGB-D and NYU Depth V2. Fully reproducible.
Go Java Python C/C++ SQL · PyTorch FastAPI Spring Boot PostgreSQL (pgvector) Neo4j · AWS Docker CI/CD