🔭 Currently: AI Engineer at Veritex Bank (Dallas, TX) — building production agentic AI platforms with LangChain, LangGraph, and Semantic Kernel, RAG/GraphRAG pipelines, LLM inference endpoints via AWS Bedrock & SageMaker, and FastAPI microservices powering high-throughput financial AI workflows.
🤝 Looking to collaborate on: Real-world AI systems with meaningful production constraints — agentic workflows, RAG/GraphRAG architectures, LLM fine-tuning (LoRA/PEFT), and MLOps pipelines. If it ships to prod and handles failure gracefully, I'm interested.
🌱 Currently deepening: Advanced multi-agent orchestration with LangGraph, GraphRAG with Neo4j, and cross-cloud model hosting strategies across AWS Bedrock, Azure AI Services, and GCP.
💬 Ask me about:
- Agentic AI platforms: multi-agent workflows, MCP servers, tool schema design, LLM gateways
- RAG & GraphRAG pipelines: FAISS, ChromaDB, Pinecone, Weaviate, Neo4j knowledge graphs
- LLM deployment & cost optimization: AWS Bedrock/SageMaker, LoRA/PEFT fine-tuning, Redis prompt caching
- FastAPI microservices: async patterns, JWT/RBAC, Celery, webhook integrations
- Full-stack AI: React/Next.js dashboards wired to ML backends, PySpark batch inference
⚡ Fun fact: I've shipped AI systems in banking and healthcare — domains where model failures have real consequences. That keeps the bar high and the excuses low.
Ramya, P., Tummala, H. C., et al. (2023). Number Plate Recognition Using Optical Character Recognition and Connected Component Analysis. Smart Technologies in Data Science and Communication, Lecture Notes in Networks and Systems, Vol. 558, pp. 29–40. Springer, Singapore. https://doi.org/10.1007/978-981-19-6880-8_3