Hyderabad, India Email: tharunravula22@gmail.com

I am an AI Engineer with 3.5+ years of experience specializing in Generative AI, Agentic Systems, and production-grade LLM pipelines.

I design and deploy end-to-end AI systems covering:

• Data ingestion and document processing
• Embedding and retrieval optimization
• Agent-based orchestration workflows
• LLM integration and evaluation
• Scalable backend services and cloud deployment

My work focuses on building high-reliability, low-latency AI systems that solve real-world enterprise problems, particularly in document intelligence, financial workflows, and automation systems.

Currently, I am working on enterprise-grade Agentic AI systems for Accounts Payable automation, along with advanced multi-agent orchestration and RAG pipelines.

I am open to full-time, contract, and freelance opportunities in:

• AI Engineering
• LLMOps
• Agentic AI Systems
• Applied Generative AI

RAG pipelines (standard and agentic), LangChain, LangGraph, CrewAI, ReAct agents, prompt engineering, grounding validation, hallucination control, evaluation frameworks.

OpenAI (GPT-4o, GPT-4o-mini), HuggingFace Transformers, Groq APIs, Ollama.

OCR pipelines, invoice parsing, entity extraction, semantic retrieval, structured validation systems.

FAISS, hybrid retrieval, ANN indexing, ranking pipelines, semantic chunking, relevance optimization.

Python, FastAPI, Flask, REST APIs, async services, microservices architecture.

PostgreSQL, MongoDB, Kafka event streaming.

Docker, Kubernetes (EKS/GKE), Jenkins, GitHub Actions, SonarQube, Aqua Trivy.

AWS (ECS, EKS, EC2, Lambda, S3, IAM), GCP Cloud Run, Azure exposure.

Latency tracking (p50/p95), token usage monitoring, system reliability dashboards.

Dec 2025 – Present | Chennai, India

• Building an Agentic AI Accounts Payable automation platform for invoice validation against PO and GRN.
• Developed LLM-based extraction pipelines converting invoice PDFs into structured JSON using OCR + parsing.
• Designed validation agents using rule-based + fuzzy matching + tolerance logic for 3-way matching.
• Built exception analysis workflows for finance teams and vendor communication.
• Implemented SLA monitoring systems with automated escalation pipelines.
• Reduced invoice validation latency from ~600 ms to ~220 ms.
• Developed FastAPI microservices processing 1000+ invoices/day.

Oct 2023 – Dec 2025 | Bangalore, India

• Built production-grade RAG systems for enterprise document intelligence.
• Improved retrieval relevance by 18–20% using semantic chunking and embedding optimization.
• Designed scalable LLM APIs handling 2000+ daily requests on AWS ECS and GCP Cloud Run.
• Implemented CI/CD pipelines (Jenkins + Docker) with integrated security scanning.
• Built observability systems tracking latency, token usage, and reliability.
• Mentored engineers on RAG architecture and LLM evaluation.

Sep 2022 – Oct 2023

• Developed ETL pipelines and automated reporting systems.
• Ensured data quality with SQL validation checks.
• Supported AI/ML pipelines with structured data workflows.

Tech: FastAPI · LangGraph · CrewAI · GPT-4o · MongoDB · PostgreSQL · Kafka · Docker

• Built a complete AI-driven invoice processing system.
• Implemented agent-based validation workflows for PO–GRN–Invoice matching.
• Designed multi-agent pipelines for anomaly detection and exception handling.
• Achieved 1000+ invoices/day throughput with ~220 ms latency.

Tech: LangChain · HuggingFace · AWS · FAISS · Flask · Docker · Jenkins

• Built a medical knowledge retrieval chatbot using RAG.
• Reduced latency from ~350 ms to ~140 ms.
• Improved top-1 retrieval relevance by ~18%.
• Deployed with full CI/CD and monitoring integration.

Tech: LangGraph · FastAPI · Streamlit · Docker · AWS · Groq

• Developed a multi-agent orchestration system for tool execution and reasoning.
• Implemented validation pipelines and retry logic.
• Improved task success rate by ~20%.

• Built a low-latency financial document RAG system.
• Reduced response latency from ~300 ms to ~150 ms.
• Improved retrieval precision by ~20%.

• Built an embedding-based recommendation system.
• Deployed on GCP with monitoring via Grafana.
• Optimized ANN retrieval for better relevance.

• Strong ability to convert GenAI ideas into production systems
• Deep expertise in RAG architecture and retrieval optimization
• Hands-on experience with agentic workflows and orchestration
• End-to-end ownership: data → model → deployment → monitoring
• Focus on latency, scalability, and reliability improvements
• Experience in enterprise-grade AI systems and secure deployments

• AI Engineer roles
• LLMOps / AI Platform Engineering
• Agentic AI Systems
• RAG Engineering
• Freelance AI system development