Architecting
Intelligent
Systems
Architecting Intelligent Systems — delivering production-scale ML systems and agentic orchestration for manufacturing and healthcare.
Vibhor Janey
AI_SOLUTION_ARCHITECT
Bristol Myers Squibb
About
Background
Vibhor Janey is an experienced AI Solution Architect with expertise in AI solution architecture, data engineering, and machine learning applications.
Currently serving as Senior Manager of AI Solution Architect at Bristol Myers Squibb, leading development of AI copilot experiences for manufacturing operations and building advanced data architecture solutions.
Experience
My journey through the AI industry, from research to production-scale implementations
Senior Manager, AI Solution Architect, GPS Business Insights and Technology
Leading AI solution architecture for manufacturing operations, delivering copilot experiences to 5,000+ users while implementing advanced LLM observability and agentic orchestration systems.
Manager, Data Architecture, Global Product Development and Supply
Architected comprehensive data solutions for manufacturing operations, creating unified batch genealogy systems that reduced processing time by 40% and improved release decision efficiency by 50%.
Machine Learning Specialist
Developed high-accuracy computer vision solutions for laboratory automation, achieving 98.59% classification accuracy while optimizing models for edge deployment and conducting analytical insights.
Software Engineer
Led front-end development and UX design for new product launches while contributing to advanced analytics solutions including anomaly detection models and banking branch management systems.
Featured Projects
From research prototypes to production systems serving thousands
Deviation Assistant
Leading the architecture and development of an agentic AI solution targeting 5,000+ manufacturing and quality users at Bristol Myers Squibb. Features an autonomous tool-calling agent orchestration built in LangGraph/LangChain and deployed on AWS ECS.
Batch Genealogy Data Product (BGDP)
Architected a unified batch genealogy system integrating SAP, Oracle EBS, and CMO sources. Reduced data processing time by 40%+ and improved release decision efficiency by 50% through graph-based data modeling.
Microplate Classification System
Built a computer vision model to classify microplates for the FAST Liquid Handler at Formulatrix, achieving 98.59% accuracy using TensorFlow and Classification-by-Retrieval (CbR) methodology. Optimized for edge deployment on Raspberry Pi and Coral Edge TPU.
Banking Branch Management System
Contributed to the inception and design of a new banking branch management product at Zebra Technologies, focusing on UX design in Figma/Adobe XD and front-end implementation with React.
Anomaly Detection Engine
Deployed an anomaly detection model combining boxplot methodology with FBProphet for time series forecasting and outlier identification in banking analytics.
Cotton Pest Classification — Few-Shot Prototypical Networks (PyTorch)
Proposed and implemented a few-shot prototypical network to identify cotton crop pests with limited annotated samples; trained on data from Li et al., *Crop pest recognition in natural scenes using convolutional neural networks*.
Skills
A comprehensive toolkit spanning AI/ML, data architecture, and production engineering
Agentic Architecture
Programming
Cloud DevOps
Data Architecture and Knowledge Management
ML/Analytics
LLMOps
Get In Touch
Let's discuss AI opportunities, research collaborations, or challenging problems
Let's Connect
Whether you're interested in discussing cutting-edge AI research, exploring collaboration opportunities, or have a challenging problem that could benefit from AI solutions, I'd love to hear from you.
Prefer to copy email?
vibhor.janey@gmail.com
Currently accepting consulting opportunities and research collaborations