Achyuth Hs

AI/ML practitioner with hands-on experience in building secure, intelligent applications using machine learning, deep learning, generative AI and agentic AI systems. Experienced in multi-agent applications, advanced RAG-based applications, and physics-informed neural networks with a strong focus on building scalable production-grade systems. Track record of reducing inference latency, deploying cloud-native solutions, and shipping compliant AI projects that create real value for the business.

Software Developer AI Intern at ElevateTrust AI (2025-12 – Present)

• Developed a multi-tenant AI customer support SaaS (Chattiq) that enables businesses to deploy custom, production ready, asynchronous RAG-powered agents across Web and WhatsApp, using AWS SQS FIFO queues and Redis atomic credit tracking to prevent concurrent race conditions, and implementing LLM evaluation and Guardrails for high security.
• Integrated LiveKit, Vobiz(Telephony), SarvamAI, and AWS to build a real-time, ultra-low latency inbound and outbound Voice AI Agent, capable of natural spoken conversations, seamless appointment booking, and instant database synchronization in PostgreSQL.
• Built a knowledge base RAG system using LangGraph, AWS Bedrock, and Redis vector search, accelerating query response times through redis HNSW search filter logic, supporting KB-scoped document isolation, web search fallback via Firecrawl, and deploying the CI/CD pipeline.
• Collaborated in fine-tuning YOLOv8 models on a custom-annotated CCTV image dataset for real-time weapon detection, and architected an AI-powered CCTV Video Intelligence system featuring automated frame extraction, frame summarizer and a conversational "Chat-with-Video" interface using ivsion language models(VLMs), reduced inference latency by 40% on consumer hardware by implementing 4-bit quantization (GGUF) for the Qwen 2.5 model.

– VIT Bhopal University