Skip to main content

Applied AI Systems Engineer

Nithin Varghese

I turn AI prototypes into deployed systems.

Built large-scale mine video analytics, 10,000+ camera exam proctoring workflows, real-time CCTV intelligence, and multi-agent LLM products. Former Head CV Engineer now building agentic AI products with LangGraph, RAG, voice workflows, and multimodal pipelines.

10,000+ cameras
Exam proctoring analytics at national-exam scale
7 mine deployment
12 analytics use cases and 1.4M+ AI safety events
Production CV
Industrial safety, ATM security, and HPCL wagon tracking
Multi-agent products
LangGraph storybook generation with print-ready PDFs
Voice-first agents
Socratic ideation workflows for structured writing outputs

More production deployments

More production deployments.

Additional shipped analytics work across exam operations, field safety, banking security, and industrial tracking.

Scale

National Exam Proctoring Analytics

10,000+ camera-scale monitoring workflows for exam operations.

Safety

Industrial Safety Monitoring

Safety analytics for field environments, including hazard and compliance monitoring.

Security

ATM Security Analytics

Security-focused video analytics for banking environments.

Operations

HPCL Wagon Tracking

Wagon tracking and counting workflows for industrial operations.

Where I can ship immediately

I build the layer between model output and real users.

The throughline is production judgment: state, latency, artifacts, monitoring surfaces, field hardware, and interfaces that make AI behavior usable.

Agentic systems

Multi-agent workflows with product-shaped outputs

I build agent loops around real artifacts: files, PDFs, voice flows, RAG contexts, tools, and stateful workspaces.

LLM systems

The layer between model output and real users

LLM work spans RAG, fine-tuning patterns, structured outputs, guardrails, workflow design, and product UX.

Production computer vision

Realtime analytics that survive field deployment

I have shipped smart CCTV analytics, industrial monitoring, and edge inference across RPi, CPU, and GPU targets.

AI engineering stack

Tools for products, pipelines, and deployment.

LangGraphRAGLLM agentsMulti-agent workflowsFine-tuningPyTorchTensorFlowTensorRTTFLiteOpenVINOEdge deploymentRaspberry PiCPU/GPU inferenceVideo analyticsObject detectionOCR pipelinesVector DB retrievalFacial recognition