Looks like you have JavaScript disabled. For the full Twine experience, you will need to re-enable it.

Artificial Intelligence Engineer with 3+ years of experience architecting high-throughput, low-latency GenAI and predictive ML pipelines for financial and enterprise systems. Expertise includes optimizing LLM inference (vLLM, TensorRT-LLM) and deploying agentic workflows (LangGraph, CrewAI) alongside advanced RAG architectures (GraphRAG, Pinecone) on Kubernetes and AWS. Track record of delivering production systems with strong reliability and observability, including hybrid semantic-graph retrieval across large-scale datasets, GPU/memory optimization, and end-to-end ML lifecycle management with monitoring and experiment tracking.…Artificial Intelligence Engineer with 3+ years of experience architecting high-throughput, low-latency GenAI and predictive ML pipelines for financial and enterprise systems. Expertise includes optimizing LLM inference (vLLM, TensorRT-LLM) and deploying agentic workflows (LangGraph, CrewAI) alongside advanced RAG architectures (GraphRAG, Pinecone) on Kubernetes and AWS. Track record of delivering production systems with strong reliability and observability, including hybrid semantic-graph retrieval across large-scale datasets, GPU/memory optimization, and end-to-end ML lifecycle management with monitoring and experiment tracking.

Deekshith Ramavath

AI Engineer, Data Scientist, Software Engineer, +2





Artificial Intelligence Engineer with 3+ years of experience architecting high-throughput, low-latency GenAI and predictive ML pipelines for financial and enterprise systems. Expertise includes optimizing LLM inference (vLLM, TensorRT-LLM) and deploying agentic workflows (LangGraph, CrewAI) alongside advanced RAG architectures (GraphRAG, Pinecone) on Kubernetes and AWS. Track record of delivering production systems with strong reliability and observability, including hybrid semantic-graph retrieval across large-scale datasets, GPU/memory optimization, and end-to-end ML lifecycle management with monitoring and experiment tracking.…Artificial Intelligence Engineer with 3+ years of experience architecting high-throughput, low-latency GenAI and predictive ML pipelines for financial and enterprise systems. Expertise includes optimizing LLM inference (vLLM, TensorRT-LLM) and deploying agentic workflows (LangGraph, CrewAI) alongside advanced RAG architectures (GraphRAG, Pinecone) on Kubernetes and AWS. Track record of delivering production systems with strong reliability and observability, including hybrid semantic-graph retrieval across large-scale datasets, GPU/memory optimization, and end-to-end ML lifecycle management with monitoring and experiment tracking.

Available to hire

Artificial Intelligence Engineer with 3+ years of experience architecting high-throughput, low-latency GenAI and predictive ML pipelines for financial and enterprise systems. Expertise includes optimizing LLM inference (vLLM, TensorRT-LLM) and deploying agentic workflows (LangGraph, CrewAI) alongside advanced RAG architectures (GraphRAG, Pinecone) on Kubernetes and AWS.

Track record of delivering production systems with strong reliability and observability, including hybrid semantic-graph retrieval across large-scale datasets, GPU/memory optimization, and end-to-end ML lifecycle management with monitoring and experiment tracking.

Skills

Experience Level

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Expert

Language

English

Advanced

Work Experience

Artificial Intelligence (AI) Engineer at CitiGroup

September 1, 2025 - Present

Engineered an autonomous compliance-checking agent using CrewAI and Llama 3 (70B) to parse 500K+ daily legal documents, reducing manual review time by 85% with 99.2% accuracy (DeepEval). Architected a low-latency LLM serving infrastructure on AWS SageMaker using NVIDIA Triton and TensorRT-LLM with AWQ quantization, reducing latency from 1.2s to 180ms at 10,000+ QPS. Built a hybrid Semantic-GraphRAG pipeline using Neo4j and Milvus to interlink complex financial entities across 50TB of datasets, improving retrieval precision by 40% and generating actionable risk insights. Deployed model lifecycle management with Ray Serve on distributed Kubernetes, integrating LangSmith and Weights & Biases for prompt telemetry; reduced model drift by 22% and maintained 99.9% uptime for AI APIs.

Artificial Intelligence (AI) Engineer at CitiGroup, United States

September 1, 2025 - Present

Engineered an autonomous compliance-checking agent using CrewAI and Llama 3 (70B) to parse 500K+ daily legal documents, reducing manual review time by 85% and achieving 99.2% accuracy (DeepEval). Architected a low-latency LLM serving stack with NVIDIA Triton and TensorRT-LLM on AWS SageMaker, using AWQ quantization to cut inference latency from 1.2s to 180ms at 10,000+ QPS. Designed a hybrid Semantic-GraphRAG pipeline leveraging Neo4j and Milvus to interlink financial entities across 50TB of unstructured transactional data, improving retrieval precision by 40% and producing actionable risk insights. Orchestrated model lifecycle management by deploying Ray Serve on distributed Kubernetes, integrating LangSmith and W&B prompt telemetry to reduce production drift by 22% while maintaining 99.9% uptime for AI APIs.

AI Engineer at Informative Web Solutions

July 1, 2021 - December 1, 2023

Developed a churn prediction engine processing 10M+ daily user telemetry events using Kafka and Spark; trained optimized XGBoost pipelines contributing to a 15% increase in quarterly recurring revenue. Built and fine-tuned BERT-based NLP models with PyTorch and Hugging Face PEFT for automated customer support ticket routing, processing 2M+ monthly inquiries and cutting average resolution time by 60%. Engineered high-throughput defect detection using ViT/CNN models; deployed an ONNX-optimized FastAPI edge pipeline achieving 500+ images/second with 98.7% F1. Automated ML CI/CD with GitHub Actions, Docker, and MLflow on GCP Vertex AI, accelerating releases from bi-weekly to daily. Created an enterprise knowledge extraction system using semantic QA with ChromaDB and LangChain, indexing 1M+ documents to save 10 hours/week in retrieval. Implemented robust ETL pipelines with Airflow to ingest unstructured web data into Databricks Delta Lake, reducing ingestion latency by 35%.

AI Engineer at Informative Web Solutions, India

July 1, 2021 - December 1, 2023

Developed a churn prediction engine processing 10M+ daily user telemetry events with Kafka and Spark; built optimized XGBoost pipelines improving retention strategies and driving a 15% increase in quarterly recurring revenue. Built and fine-tuned BERT-variant NLP models using PyTorch and Hugging Face PEFT for automated customer support ticket routing, processing 2M+ monthly inquiries and reducing average resolution time by 60% while lowering manual triage costs. Engineered high-throughput defect detection using ViT and CNNs, deploying an ONNX-optimized FastAPI pipeline at the edge to process 500+ images/second with 98.7% F1-score. Automated ML deployment CI/CD using GitHub Actions, Docker, and MLflow on GCP Vertex AI, improving deployment frequency from bi-weekly to daily. Implemented an enterprise knowledge extraction system using semantic QA with ChromaDB and LangChain to index 1M+ legacy documents, saving teams ~10 hours/week in retrieval. Built robust ETL pipelines using Airflow to