Open to opportunities

vinat goyal

@vinatgoyal1

Message

Computer vision and AI engineer focused on production-ready ML systems.

India

Message

What I'm looking for

I seek roles building and deploying computer-vision and ML systems at scale, with emphasis on production reliability, model optimization, and collaborative, engineering-driven teams.

I am a computer vision and AI engineer with hands-on experience building and deploying production-grade models for industrial and surveillance applications.

At Strayos I led data annotation and developed a SegFormer-based haul-road segmentation pipeline, migrating workflows to orthophotos and achieving an IoU of 0.78 and F-score of 0.86.

I optimized legacy model pipelines—converting Keras models to ONNX for a 30% performance improvement—and built an RAG-based LLM pipeline with LangChain to summarise production error logs and speed debugging.

Previously I developed a weapon-detection system using YOLOv7 with >85% precision, curated multiple diverse datasets including synthetic data, and created a multimodal document-classification model (DocNet) that achieved 91% accuracy and reduced underwriter processing time by 75% via a FastAPI + Docker deployment.

Experience

Work history, roles, and key accomplishments

Current

Computer Vision & AI Engineer

Current

Strayos

Feb 2025 - Present (8 months)

Led data annotation and development of a computer vision pipeline to detect rutting, berms, and rock spillage; migrated haul-road segmentation to a SegFormer-based orthophoto workflow achieving IoU 0.78 and F-score 0.86 while optimizing rock fragmentation pipelines for 30% performance gains.

Computer Vision SegFormer Onnx PyTorch Optimization Data Annotation RAG

Computer Vision Engineer

Eagle Eye Networks

Sep 2023 - Feb 2025 (1 year 5 months)

Developed and optimized a YOLOv7-based weapon detection system achieving >85% precision and curated 10 diverse datasets including synthetic pipelines to improve model robustness and deployment readiness.

YOLOv7 Synthetic Data Model Fine Tuning PyTorch Video Surveillance Data Pipelines Model Evaluation

Data Science Intern

Munich Reinsurance

Sep 2022 - Jan 2023 (4 months)

Built DocNet, a multimodal document classification model (BERT + DiT) achieving 91% accuracy over 11 classes; deployed via FastAPI in Docker, reducing underwriter processing time by 75% and produced client analytics foundations in Power BI.

Multimodal DL BERT Hugging Face fastAPI Docker Data Cleaning Power BI

Education

Degrees, certifications, and relevant coursework

Indian Institute of Information Technology

Bachelor of Technology, Computer Science and Engineering

2019 - 2023

Completed a Bachelor of Technology in Computer Science and Engineering with coursework and projects focused on computer vision, deep learning, and software deployment.