Open to opportunities

Harish Nagari

@harishnagari

Message

AI/ML Engineer at INAI – iHub, IIIT Hyderabad building road-asset detection and RAG pipelines for real-world data.

India

Message

At INAI – iHub, IIIT Hyderabad, I built and trained YOLOv8/v11 road-asset detection models for real-world imagery and developed inference pipelines calibrated to reduce false positives. I also created a RAG pipeline for metadata retrieval and Grafana dashboards to monitor model performance and pipeline health.

At Scale AI (Outlier), I designed evaluation frameworks for AI model outputs and curated training datasets across NLP and computer vision. Earlier, at XenonStack, I developed a fraud detection system and AWS-based ETL pipelines; my projects include real-time data processing with PySpark Structured Streaming and an emotion and age detection system.

Experience

Work history, roles, and key accomplishments

Current

Machine Learning Engineer – Associate

Current

INAI – iHub, IIIT Hyderabad

Feb 2026 - Present (8 months)

Built and trained deep learning road-asset detection models and binary classification models for defect and anomaly detection. Developed RAG pipelines, Grafana dashboards, and digital twins, and managed model evaluation with MLflow.

YOLOv8 YOLOv11 PyTorch Deep Learning Object Detection Classification Anomaly Detection RAG Embeddings Vector Databases Grafana MLFlow Voxel51 Data Pipelines Mentorship

AI Data Engineer

Scale AI (Outlier)

Nov 2024 - Dec 2025 (1 year 1 month)

Designed structured evaluation frameworks for AI model outputs and worked with graph data and vector databases for semantic retrieval. Curated and annotated large-scale training datasets across NLP and computer vision domains.

AI Evaluation Graph Databases Cypher Vector Databases Semantic Retrieval NLP Computer Vision Data Annotation Distributed Team Collaboration

Associate Software Engineer Trainee

XenonStack

Feb 2024 - Aug 2024 (6 months)

Developed a fraud detection system using scikit-learn and TensorFlow, architected ETL pipelines on AWS, and built distributed real-time data pipelines with Spark, Kafka, and Cassandra. Deployed containerized services using Docker and Kubernetes.

scikit learn TensorFlow Fraud Detection ETL AWS S3 Athena Glue Redshift Kinesis SQL Spark Kafka Cassandra Docker Kubernetes

Artificial Intelligence Intern

UpSkillz

May 2022 - Sep 2022 (4 months)

Implemented and trained CNNs for image classification, improving accuracy by 20% through architecture tuning and hyperparameter optimization. Built end-to-end Python ML pipelines and presented results to senior leadership.