Experience

A timeline of research collaborations and industry roles focused on applied AI, LLM systems, and enterprise-grade automation.

March 2025 – August 2025

AI Intern

BrandContext.ai • Hybrid, Mumbai, India

• Developed an AI-agentic system powered by dynamic graphs and real-time factory operation databases to support supervisor workflows, detect anomalies, and dynamically incorporate worker-input special conditions.
• Evaluated and productionized computer-vision pipelines for retail supermarkets, fine-tuning YOLOv11 multi-class detectors for Xiaomi (0.87 mAP@50 across over 500 stores in India).
• Built legally-compliant speaker-diarization pipelines for sales–customer calls using open-source (SpeechBrain) and commercial (ElevenLabs) stacks to enable accurate speaker labeling and downstream analytics.

October 2024 – May 2025

Research Intern

Indian Institute of Technology, Patna • Remote

• Designed MetaSearch, a search-augmented, reasoning-based AI Agent for consensus resolution in peer review, combining disagreement detection and fact-grounded synthesis to automate meta-review decision-making.
• Developed a novel LLM-argumentation system for peer review, evaluating 900+ ICLR reviews using leading LLMs (Llama 3.1, Mixtral, Gemini 2.0, GPT-4o), achieving high inter-annotator agreement (Cohen's Kappa = 0.834).

August 2024 – February 2025

AI Intern

Nvelop Technologies Oy • Remote, Helsinki, Finland

• Devised Retrieval-Augmented Generation (RAG) workflows in Azure AI PromptFlow across 50+ RFPs.
• Optimized chunking strategies and top-p sampling parameters, reducing document retrieval perplexity by 15%.

June 2024 – July 2024

Research Assistant

UIUC+ Summer Research Program • USA (Remote)

Investigated C-to-Rust transpilation with existing tools and LLMs, categorising translation bugs and strengthening quality evaluation frameworks for future research.

June 2024 – July 2024

Python (LLM) Development Intern

PavePilot AI • USA (Remote)

Built AI-powered marketing content pipelines, evaluated multiple image-generation models, and tailored GenAI prompts to create platform-specific, brand-aligned campaigns.