Hi, I'm Jay Shah,an ML engineer.
Senior Machine Learning Engineer at 6sense, building production AI applications, foundational embeddings, and agent evaluation frameworks.
About Me
I build machine-learning and language-model systems that have to work beyond the notebook.
I write about the systems I build, the assumptions behind them, and the parts that usually break. If something is on this site, it shipped or I learned enough from it to explain what happened.
Career & Engineering Timeline
Senior Machine Learning Engineer • 6sense
Leading AI systems engineering initiatives for enterprise intent intelligence. Architecting foundational models with custom embeddings, model explainability pipelines, contextual knowledge graphs, and evaluation harnesses for autonomous agents.
- ›Trained foundational models with custom embeddings to power B2B intent intelligence at enterprise scale.
- ›Shipped model explainability in production to provide transparent predictions for sales and marketing workflows.
- ›Architected a LLM agent evaluation framework adopted across the engineering and product organizations to prevent regressions.
- ›Built internal agents for key revenue acceleration use cases.
- ›Designed and developed a contextual knowledge graph to link enterprise accounts, intent signals, and buyer personas.
Data Scientist III • Avathon
Led foundation-model MLOps platforms, architected agent platforms with Model Context Protocol (MCP) tools, and built production anomaly detection and renewable energy forecasting systems across enterprise deployments.
- ›Led foundation-model MLOps platform; cut release time 45% and increased model adoption.
- ›Architected agent platform with MCP tools to spin up domain agents in under 5 minutes; powered 20+ workflows.
- ›Built LLM interface integrating asset data, enabling reporting and task automation.
- ›Launched cross-site solar-storage autoencoder anomaly detection to predict failures and performance issues, saving >$500k.
- ›Shipped RAG compliance agent, reducing violations 10% and automating audit preparation for energy domains.
- ›Reduced false positives 25% by ranking predictive alerts and next-best actions with Bayesian analysis.
Graduate Research Assistant • Texas A&M University
Graduate research focused on applying advanced machine learning and deep learning methods to solve and predict wind energy system failures under Dr. Yu Ding.
- ›Researched with Dr. Yu Ding on applying advanced machine learning methods to solve and predict wind energy system failures.
- ›Implemented deep learning methods to predict possible power production and downtimes associated with wind turbine failures.
- ›Awarded Outstanding Master of Science Student (Apr 2019) by Department of Industrial and Systems Engineering.
Outside the Terminal
Cricket comes first. Tea, never coffee. Yoga is the habit that survived a newborn's sleep schedule.
I care about Indic languages and culture, especially the gap between the languages people speak and the languages most AI systems understand. Gujarati Llama started as a personal project because useful language technology should not be limited to English-speaking users.
I also read outside machine learning: systems thinking, philosophy, and energy policy. Some of my best technical ideas began as questions from another field.
Honors & Awards
Developed StreamLens, an agentic multi-model RAG system for interacting with autonomous vehicle video streams.
Texas A&M University Department of Industrial and Systems Engineering.
Patents
Filed Dec 30, 2021. Systems and methods for computing operational losses across electrical grid outages.
Filed Dec 30, 2021. Machine learning models forecasting multi-horizon generation profiles.
Filed Dec 11, 2017. Automated dimensional calibration and defect separation via computer vision.