• 🎓 B.Tech CSE Final Year
• 🔬 Research Intern @ IIIT Hyderabad (Neural Rendering, 3D Vision)
• 🧪 Focused on 3D Computer Vision and Neural Rendering
• ⚡ Interested in bridging research → real-world AI systems

• Developed a CLIP-guided 3D Gaussian Splatting framework integrating custom opacity regularization and dynamic point cloud pruning.
• Integrated Vision-Language supervision to improve semantic scene understanding and eliminate transient reconstruction artifacts.
• Optimized multi-view 3D reconstruction modules to stabilize novel view synthesis within dynamic, non-static environments.

• Profiled and evaluated key foundational NeRF variants (Instant-NGP, SSDNeRF, RobustNeRF) across complex localized datasets.
• Optimized rendering pipelines to improve convergence speed, balancing real-time frame-rate throughput against structural Peak Signal-to-Noise Ratio (PSNR) metrics.

A. Prabakaran, et al., "Targeted Data Augmentation for Hallucination Mitigation and Faithfulness Enhancement in Compact LLMs," Proceedings of the IEEE INDIACOM Conference, 2026

Core Contribution: Developed an experimental framework utilizing parameter-efficient fine-tuning (PEFT) and calibration optimization to evaluate factual faithfulness metrics and quantify model confidence boundaries.

• Developed a QMIX-based cooperative multi-agent reinforcement learning framework for dynamic delivery routing under traffic, weather, rider health, and delivery urgency constraints.
• Integrated Yen’s K-Shortest Paths, real-world road networks, and flood-prioritized experience replay to improve robustness under adverse conditions.
• Evaluated the system across multiple city environments using delivery cost, SLA compliance, route efficiency, and fleet coordination metrics.

• Proposed a targeted data augmentation strategy to reduce hallucinations and improve faithfulness in compact language models.
• Designed experiments around failure-specific augmentation and controlled evaluation to measure factual consistency and reliability improvements.
• Published the work at INDIACom 2026 (IEEE) as “Targeted Data Augmentation for Hallucination Mitigation and Faithfulness Enhancement in Compact LLMs.”

• 🎓 IASc Summer Research Fellowship Recipient — National fellowship awarded by the Indian Academy of Sciences (IAS); selection rate historically under 2%.
  
  
• 🏅 Smart India Hackathon 2025 | National Finalist (Top 5) — Selected as a Top 5 finalist from over 50,000 national applicants in the Ministry of Education's flagship engineering competition.
  
  
• 🌏 Google AI for Impact Hackathon - Selected into the final cohort of 98 teams out of thousands of competing teams across the Asia-Pacific (APAC) region.
  
  

⚡ Open to research collaborations, internships, and AI-focused opportunities