nobody User

Building analytics capabilities over accounting and financial data, surfacing insights from transactional records to support finance workflows. Developing conversational AI features on Anthropic's Claude API, enabling users to query accounting data in natural language.

Built Brain OS, the LLM reasoning and task-orchestration layer for a fleet of warehouse robots running on ROS, interpreting operational requests and routing tasks to appropriate agents. Designed a multi-model architecture orchestrating GPT, LLaMA, and Mistral with BERT-based intent classifiers, and deployed the system on Azure.

Built the dataset export pipeline converting a 100,000+ image corpus into YOLO training format, and shipped perimeter breach detection and people-count processing into EVA with real-time inference across 8–16 concurrent camera streams. Developed an in-house image annotation tool and integrated Kafka-based streaming with Flask REST APIs.

Implemented Federated Learning and Gossip Learning protocols for distributed model training across simulated multi-node setups, benchmarking convergence and accuracy against a centralised baseline. Researched image processing techniques for distributed systems and benchmarked model efficiency and scalability trade-offs.

Grade: 8.9/10

Master of Computer Applications (MCA) from SRM Institute of Science and Technology, Chennai, India, from June 2021 to June 2023.

Grade: 7.4/10

Bachelor of Science in Computer Science from The New College, Chennai, India, from May 2018 to May 2021.

Certification in Data Analytics, Artificial Intelligence & Machine Learning from Besant Technologies, completed in June 2023.