cv

Research Engineer — Algorithms and Applications

• Developing AI and quantum algorithms for scientific computing, with a current focus on solving partial differential equations.

Research Intern — Advanced Reasoning

• Built a framework for generating synthetic mock MCP servers and large-scale tool-calling datasets.
• Worked on reinforcement-learning post-training of small language models for complex tool-use workflows.
• Designed rubric-based reward systems for training and evaluating deep-research agents.

Quantum Research Intern — AI for Quantum

• Built a CLI coding agent to assist with translating Python components of Qiskit workflows into Rust.
• Contributed agentic code generation and context-aware assistance to Qiskit developer tooling.
• Developed retrieval-augmented assistants for quantum-computing workflows and support.

AI Research Intern

• Developed a scalable synthetic-data framework for adapting vision-language models to multilingual documents across 22 languages.
• Helped create the Nayana dataset with millions of document images for OCR, VQA, layout analysis, and retrieval.
• Trained and released models and co-authored work at NAACL, CVPR, ICCV, and related workshops.
• The project received a $100,000 Meta Llama Impact Grant.

Summer Research Intern

• Developed Parichaya, a retrieval and multimodal question-answering interface for colloquial Kannada community knowledge.
• Built the supporting application with Next.js and FastAPI under Prof. Srinath Srinivasa.

B.Tech. in Computer Science and Engineering

• CGPA 8.5/10

ReviewArena

• Built a cross-conference dataset of 51,529 papers and 196,099 reviews and a venue-aware benchmark for LLM peer review.
• Accepted as a Spotlight at the ICML 2026 AI for Science workshop.

HyperCluster

• Developed decentralized LLM inference across heterogeneous devices using peer-to-peer networking, automatic model sharding, and ring-based pipelining.
• Published at DAIS 2026 and presented at the AAAI 2026 ML4Wireless workshop.

manimator

• Built an open-source system that transforms research papers and mathematical concepts into Manim animations using LLMs.
• Reached more than 15,000 uses in its launch week and was featured as a Hugging Face Space of the Week.

Nayana and SynthDoc

• Built multilingual synthetic-document generation infrastructure for OCR, VQA, layout analysis, and retrieval across 22 languages.
• Scaled dataset generation across 100 CPU and 10 GPU workers.

KissanDial

• Developed a voice-call AI assistant for farmers using LlamaIndex, Twilio, FastAPI, and Milvus.

• Spotlight paper and contributed talk, ICML 2026 AI for Science Workshop — ReviewArena

• Meta Llama Impact Grant ($100,000) — Nayana

• AMD Kria KR260 FPGA — AMD Pervasive AI Hackathon, Robotics AI category