Deepankan Srivastava

I am an AI Engineer with 2+ years of hands-on experience building LLM-driven, agentic, and multimodal Retrieval-Augmented Generation systems for enterprise customers. I design and deploy scalable, secure AI platforms—ranging from Perplexity-style private RAG engines to voice-enabled Twilio agents—while focusing on performance, reliability, and cost-effective model routing.

My work includes architecting multimodal RAG with LangGraph, integrating Llama-family models with vector stores like Pinecone and Chroma, and productionizing pipelines on AWS and GCP. I have delivered concrete outcomes such as reducing manual scheduling by 40%, cutting query resolution time by 30%, and boosting legal-document review efficiency by 50%.

I combine deep technical skills (LangChain, embeddings, fine-tuning, PyTorch/TensorFlow) with strong MLOps and cloud deployment experience to move prototypes into production. I enjoy solving domain-specific retrieval problems, building robust APIs, and optimizing systems for large-scale, secure access to organizational knowledge.