About Me
As a Lead Data Scientist and Machine Learning Engineer with over five years of experience, I specialize in building and deploying complete, end-to-end ML solutions in production cloud environments. My passion lies in translating complex enterprise problems into impactful, data-driven applications, spanning advanced NLP pipelines, decision support systems, and large-scale model deployment infrastructure. A core part of my work involves not just building models, but also the surrounding ecosystem. I have hands-on experience in microservice architecture using Java and Spring Boot to build scalable, robust backends for ML applications. I’m also deeply involved in the MLOps lifecycle, defining best practices for model monitoring, feedback loops, data and model version tracking, and CI/CD integration to ensure models perform reliably in production.
Key Projects
Custom Intent Detection Microservice
I engineered a production-scale Custom Intent Detection microservice that enables customers to train models on their own data. The system oversees 10,000+ live custom models handling approximately 10 million daily requests - one of the highest-scale ML deployments I’ve worked on end-to-end.
Root Cause Analysis (RCA) — Decision Support System
I built and deployed an RCA product feature that combines Generative AI and data analysis techniques to proactively alert business leaders about anomalies in their key business metrics. The system surfaces AI-generated potential root causes alongside the anomaly signals, turning raw data into actionable intelligence. I’m currently spearheading enhancements to support higher customer data volumes and improved anomaly detection throughput.
Query-to-Graph Agentic System
I designed and implemented a Query-to-Graph agentic system that enables customers to build analytical dashboards and conduct data-driven conversations using natural language queries, bringing semantic parsing and NLP to structured business data.
LLM Migration & Evaluation Framework
I directed the transition of production features from proprietary LLMs to open-source LLaMA and Qwen models, applying quantization techniques and evaluating output quality using MLflow’s LLM-as-a-Judge framework. I’m also architecting an organization-wide evaluation framework for agent-based AI features, standardizing quality measurement across agentic workflows.
Other Professional Experience
- Embedding Service: Designed and deployed a production-grade embedding service by quantizing the LaBSE model and building a high-performance inference layer for large-scale use.
- Entity Extraction: Built a service to identify and extract structured entities from support tickets and live chat conversations. MLOps Platform: Designed an organization-wide MLOps framework covering model monitoring, feedback loops, and data and model version tracking.
- Conversational AI: Developed and integrated a conversational AI model for small talk within the chatbot framework, and implemented Global Feedback Flows to drive continuous product improvement.