Experience

Building an Agentic AI Copilot that translates natural-language app monetization goals into validated, production-ready AdMob configurations

Building Dynamic Ledger, a test-time learning framework enabling LLMs to accumulate and reuse problem-solving strategies without fine-tuning; raised $5,000+ in funding for API compute and conducted billion-token-scale experiments

Training a Qwen3-1.7B summarizer with PPO-based RLHF alongside a LoRA-fine-tuned classifier to generate behavioral annotator profiles from rating history and predict personalized text preferences

Developed production-scale AI agents and RAG pipelines to deliver insights from large-scale financial data; authored a Java/TypeScript GenAI template library adopted by several teams

Engineered a GPU-efficient NLP pipeline with OpenAI Whisper and quantized Qwen3-14B LLM to process 2,600+ student assessments, cutting review time by 85%+ while preserving privacy

Launched 3 educational iOS apps and 2 books, generating 10k+ downloads and $15k+ sales