Anirudhan Kasthuri

• Architected high-concurrency Voice AI Agents and LLMOps using LiveKit and Docker; reduced end-to-end latency by 25% via optimized WebSocket streaming and asynchronous task queues. • Engineered production-grade RAG workflows with pgvector and LangChain; implemented sophisticated prompt chaining to eliminate hallucinations in real-time customer-facing inference.

• Fine-tuned ViT models for open-vocabulary detection; deployed models on low compute devices using quantization for significant inference speedups. • Managed Azure NC-series GPU clusters for distributed training.

• Independently built a full computer vision system from scratch, including data collection, annotation, and fine-tuning of a YOLOv5 model for detecting physically disabled individuals in hospital settings. • Deployed the optimized model using Python, OpenCV, and PyTorch for real-time inference on edge devices, improving detection accuracy by 10% under challenging lighting and occlusion conditions.

• Assisted in day-to-day manufacturing and assembly operations of electric motor applications, including drilling machines and power tools, gaining hands-on experience with industrial equipment and workflows. • Studied PLC-controlled production lines and contributed to optimizing manufacturing processes using XGBoost-based machine learning models for predictive maintenance and process improvement.

I had the privilege of attending an internship at the National Small Industries Corporation (NSIC), Chennai. The internship was held for 5 days from 13.06.2022 to 17.06.2022. The internship attended at “National Small Industries Corporation” focused on learning the basics of electrical, electronics, and embedded design. We also learned about the basics of home automation using the concept of Internet of Things (IoT). Everyone was given an insight about a circuit that converts the AC current from the t