Student

Vicky KumarDT

Thinking Engineer.

Ask me about...

LLM Infrastructure & Fine-tuning

How do I fine-tune an LLM on proprietary data safely? →What infrastructure do I need to serve my own LLM? →When should I fine-tune versus use retrieval-augmented generation? →

AI Observability & Reliability

How do I detect when my LLM starts producing bad outputs? →What metrics actually matter for monitoring LLM performance? →How do I debug unpredictable behavior in production AI systems? →

Model Internals & Circuit Discovery

How do I understand why my model makes specific decisions? →Can circuit analysis help me make my model more reliable? →What does mechanistic interpretability mean for my AI product? →