Turing
Data Scientist
- Designed prompts, agent workflows, and tool-calling strategies for Penguin's agentic AI systems using SFT and RLHF.
- Built hybrid code RAG for Amazon SWE-Bench GitHub issue analysis with Python, Neo4j, Qdrant, BM25, and cross-encoder reranking.
- Implemented LangGraph multi-hop retrieval over calls, imports, inheritance, and dependencies for grounded engineering Q&A.
- Managed a 6-person team working on graph-based data representations for large-scale learning pipelines.