AI/LLM Engineer | RAG Systems | Agentic AI | Production ML

📍 Bengaluru, Karnataka, India
📧 irakamsivabhanuprakash@gmail.com
🔗 LinkedIn: https://www.linkedin.com/in/siva-venkata-bhanu-prakash/
🔗 GitHub: https://github.com/hynko431

🔭 Currently working as an AI Engineer Intern at ASAR IT Technologies, Bangalore, where I build and deploy production-grade AI solutions including RAG systems and deterministic LLM pipelines.

👯 Open to collaborating on open-source AI and LLM projects, especially in areas involving retrieval systems, agentic workflows, and scalable API architectures.

🤝 Actively strengthening my foundation in Data Structures & Algorithms (DSA) to enhance problem-solving, system design, and performance optimization skills.

🌱 Currently advancing my expertise in Agentic AI & Generative AI technologies, focusing on real-world deployment, evaluation-driven development, and low-latency inference systems.

💬 Ask me about RAG architectures, LLM pipelines, FastAPI deployments, semantic search, and production AI engineering.

⚡ Fun fact: I believe consistent progress — especially during challenging times — builds long-term resilience and success.

ASAR IT Technologies — Bengaluru
Jan 2026 – Present

• Built deterministic LLM extraction pipelines with 99% JSON-valid outputs
• Maintained ~300ms inference latency
• Reduced OCR noise by 80%+
• Improved extraction F1 score by 18%
• Implemented RAG pipelines using LangChain + FAISS
• Developed FastAPI-based inference & retrieval APIs
• Built evaluation dashboards tracking precision, recall, F1 & latency

• Multi-agent financial advisor for 100+ MSMEs
• Reduced manual analysis time by 60%
• Integrated Shopify, Mailchat, Supabase APIs
• Built during GDG Hyderabad Agent-A-Thon

Live: https://agent-a-thon.vercel.app/
Repo: https://github.com/hynko431/Agent-A-Thon/

• Conversational medical-report RAG platform
• FAISS-based semantic retrieval
• Structured output validation using Pydantic
• Agentic orchestration pipelines

Live: https://ai-agentic-medicalreport-analysis.streamlit.app/
Repo: https://github.com/hynko431/Agentic-MedicalReport-Analysis

• Multimodal AI avatar with speech-to-text & TTS
• Integrated agentic memory and real-time interaction
• FastAPI + React architecture

Repo: https://github.com/hynko431/Conversational-AI-Avatar

• Production-grade RAG systems
• Agentic AI architectures
• Low-latency LLM inference
• Evaluation-driven ML development
• Scalable FastAPI deployments

Build real systems. Measure performance. Improve continuously.