AI Engineer &Deep Learning Researcher

Aarvian DevsarTech

Remote: Anywhere on Earth

• Led the design and implementation of RIK, a multi-agent horizontal document-intelligence platform and SDK built with LangGraph, integrating standard and heuristic RAG, generative modeling utilities, and calculation-engine tooling for multimodal ingestion and decision support.
• Developed a generative AI narrative engine for Suspicious Transaction Reports (STRs), automating AML documentation workflows through domain-specific LLM context profiles aligned with financial reporting standards.
• Architected secure data streaming pipelines (Data Faucet) and production inference workflows with LLM caching, prompt optimization, and citation indexing to deliver fact-grounded, low-hallucination outputs.

Saptkrishi Scientific Ltd.

IIT Kanpur, Kanpur, UP, India

• Designed and evaluated AI frameworks combining federated learning and edge computing for IoT devices
• Prepared reports and presentations summarizing research findings on distributed inference systems for low-power devices
• Designed marketing graphics for the Scarlet Blinks product

H2LooP.ai Inc.

San Francisco, CA, USA

• Researched Semantic trees to produce an analytical modeling of concrete, abstract and lossless semantics trees, resulting in better decision making to select generative code models for system engineering workflows; enhanced context integrity by 15%
• Developed scalable pipeline & SDK using unstructured.io to structure unorganized data in JSON with 90% accuracy, enhancing handling by 50% and reducing training data pre-processing time by 35%