Fireworks AI · Incoming Applied Machine Learning Engineer Intern
Joining the Applied ML team to enable new model stacks and optimize inference performance across production workloads.
Experience
Current
UC Berkeley EECS & Statistics · Researcher
Researching adaptive evaluation for LLM agents with Ian Waudby-Smith and Keegan Harris. Designed a framework that reduced benchmarking cost by 63% and extended it to live, non-fixed evaluation sets using a zooming-bandit formulation.
Previous
Prologis · Applied Machine Learning Intern
Built an enterprise RAG platform spanning live Dataiku, Snowflake, and OpenAI workflows for citation-backed financial analysis. Combined structured and unstructured retrieval across two engines and consolidated 1,000+ documents, cutting information-retrieval time by 85%.
Zendesk · Applied AI Engineer (Contract)
Designed and benchmarked HDAT and F-CoT ITSM-agent architectures for uncertain tasks, reaching 95% reliability. Evaluated 100+ tickets and developed evidence-based capability firewalls that reduced hallucinations and unsafe tool use by about 27%.
Breefly · Co-Founder & Head of Product
Built multi-stage LLM analysis for long-form lecture video, with sub-two-second transcript retrieval, Redis caching, Celery orchestration, and WebSocket streams. Reached 1.5K+ users and 5M+ impressions, raising a $150K pre-seed from Character Capital at a $3M valuation.
Astera Holdings · Technical Operations & Analytics Intern
Built KPI dashboards tracking seven-plus employee performance metrics, improving operational efficiency by about 60%. Also automated market reports and operational processes with agentic AI workflows, reducing reporting time by about 75%.