GenAI Engineer: Agentic LLM Systems & APIs

DEEPAK BHUVANAGIRI

GenAI Engineer · Python, LLM Systems & Agentic AI

+91-979******* · ******.********@*****.*** · linkedin.com/in/deepak-bhuvanagiri-68567877 · India SUMMARY

Python & GenAI Engineer with 9+ years of experience building production AI systems. Designs agentic workflows with LangChain and LangGraph, RAG pipelines, and FastAPI microservices serving both self-hosted and API-based LLMs. Hands-on with LLM guardrails, prompt-injection and PII defence, vector search, asynchronous orchestration, and Kubernetes delivery on AWS. Consistently turns AI prototypes into reliable, observable, production-grade services. EXPERIENCE

GenAI Developer

Tata Consultancy Services (TCS)

09/2025 - Present India

Building agentic AI and LLM platform services for regulated enterprise workflows. Built a FastAPI asynchronous orchestration service executing multi-agent LangGraph workflows, assembling agent graphs dynamically from runtime configuration so new business processes ship without API code changes.

Engineered 25+ reusable agent hooks spanning intent identification, document extraction, sentiment analysis, work-type routing, draft-response generation, and human-in-the-loop verification.

Implemented stateful resume and recovery: workflow state is persisted per service request and rehydrated to restart interrupted runs from the exact failed step, removing full reprocessing. Wrapped every agent node with guardrail, security, and supervisor layers that detect violations and auto-escalate for human review.

Protected throughput with asyncio semaphore concurrency control, background task execution, HTTP 429 backpressure, and pre-flight health gating across 6+ downstream microservices.

Delivered an LLM gateway exposing chat, NDJSON token streaming, embedding, and guardrail endpoints over self-hosted Gemma 4 (including gemma4_26b), Llama 3.2, Llama Guard 3, and nomic-embed models.

Benchmarked and tuned open-weight models across the Gemma 4 and Llama families, matching model size to latency, cost, and accuracy targets per workflow. Designed a layered guardrail engine combining content safety, jailbreak and prompt-injection detection, SQL-injection screening, and PII protection using 40+ normalized detectors with header-driven policy modes.

Built RAG chatbots and retrieval pipelines over 5,000+ PDFs at 92% relevance accuracy and 30 documents/sec, cutting manual support effort by 40%. Shipped to AWS EKS and Lambda through Podman, Helm, and Jenkins CI/CD with correlation- ID tracing, ECS-format rolling logs, and pytest regression suites. Python Developer

Tata Consultancy Services (TCS)

02/2019 - 08/2025 India

Improved operational efficiency across legacy applications, data workflows, and insurance modules.

Developed Python solutions for legacy applications, improving testing efficiency by 30%. Executed complex SQL operations and stored procedures to strengthen ETL processing and data integrity.

Architected ETL pipelines with Git, Jira, and Jupyter, increasing processing speed by 40%. Created automation frameworks for insurance modules and folded user feedback into continuous improvements.

Network & Infrastructure Support Engineer

Tata Consultancy Services (TCS)

02/2016 - 01/2019 India

Delivered network and infrastructure support sustaining 99.9% uptime against tracked SLAs. Led requirement gathering, testing, and infrastructure enhancement rollouts for enterprise users.

EDUCATION

Bachelor of Engineering (B.E.)

St. Peter's College of Engineering and Technology, Anna University 08/2011 - 05/2015 India

CORE SKILLS

GenAI & LLM: LangChain, LangGraph, RAG, Agentic

Workflows, Prompt Engineering, GPT APIs, Ollama, Gemma 4

(gemma4_26b), Llama 3.x, Llama Guard, Embeddings,

ChromaDB, FAISS, Pinecone, LLM Guardrails, PII & Jailbreak Defence

Backend: Python, FastAPI, Asyncio, Pydantic, REST APIs, Microservices, SQL, pytest

Data: Pandas, NumPy, Scikit-learn, XGBoost, ETL Pipelines, PyMuPDF, PyPDF2

Cloud & DevOps: AWS (EKS, Lambda, API Gateway, S3), Azure, Docker, Podman, Kubernetes, Helm, Jenkins, GitHub Actions, Git, Jira

KEY PROJECTS

Agentic Workflow Orchestration Service

Configuration-driven LangGraph engine running multi-step agent workflows asynchronously, with persisted state, step- level resume, guardrail and supervisor wrappers, rule-based routing, and semaphore concurrency control.

LLM Gateway & Guardrail Service

FastAPI gateway over self-hosted Gemma 4 and Llama models exposing chat, streaming, embedding, and guardrail APIs with layered content-safety, prompt-injection, SQL- injection, and PII protection, deployed on EKS and Lambda. CERTIFICATIONS

AWS Solutions Architect - Associate

PGP in Data Science and Business Analytics

Great Learning - Great Lakes Executive Learning

ITIL v3 Foundation

HIGHLIGHTS

Career transition: Moved from infrastructure support into Python and GenAI engineering through hands-on delivery and certification.

Client recognition: Repeatedly recognized by clients for timely delivery and solution quality.

Contact this candidate