Darshit Purohit

Architected and deployed a production-grade RAG platform using Qdrant, and designed a multi-stage agentic AI pipeline with function calling. Led backend architecture for a mathematical equation evaluation platform and redesigned synchronous services into asynchronous processing pipelines.

Led architecture and development of an Enterprise Knowledge Assistant for an Australian mining company, enabling natural language interaction with structured and unstructured data. Designed a multi-stage agentic AI pipeline and engineered a scalable RAG architecture using LangChain, Redis, Qdrant, and WebSockets.

Designed and developed an agentic GenAI platform with multi-source RAG, integrating web search, PDF knowledge retrieval, LangChain, Gemini, HuggingFace embeddings, Redis, and vector search. Engineered a provider-agnostic LLM architecture and developed a real-time sports betting prediction system using XGBoost, Optuna, and FastAPI.

Led research and development of an IMU-based vehicle speed estimation system using LSTM, Bi-LSTM, and Conv1D models. Designed and implemented a large-scale RAG platform over 6 million research papers using Weaviate, OpenAI, and Llama models, and led a team of three engineers.

Automated software build and deployment workflows using Python and Jenkins, reducing manual effort and deployment errors by approximately 40%. Debugged system integration issues and collaborated with cross-functional development teams.

Developed Python automation tools and contributed to CI/CD pipelines, supporting software development, testing, and deployment automation.

Developed a real-time queue detection system using Mask R-CNN and YOLOv3 on CCTV video streams, achieving 72% detection accuracy during experimental evaluation.

Pursued a Master of Science in Computer Science with a specialization in Artificial Intelligence.

Earned a Bachelor of Technology in Computer Science with a specialization in Big Data Analytics.

Completed a Diploma in Computer Engineering.