Siddhant Gupta

My path into AI engineering started with a question: how do we make language models reliable enough for production?

That curiosity took me from research labs at UMBC where I published peer-reviewed work on ML-based malware detection using YARA signatures and LightGBM to building real-world LLM systems at the Center for Applied AI. Today, I architect multi-agent pipelines, RAG implementations, and fine-tuned models that handle thousands of concurrent interactions. What excites me most is the intersection of research and production engineering. I love the challenge of taking cutting-edge techniques (structured outputs, GRPO fine-tuning, hybrid search with reranking) and making them work reliably at scale. That's why I contribute to open source projects like Meta's Llama Prompt Ops, I believe in building in public and learning with the community.

Outside of work, I write technical deep-dives on Medium and experiment with new frameworks. Always happy to discuss agentic AI, RAG optimization, or the latest in LLM tooling.