AI & Data Engineer
Production ML systemsand AI automation.
I build infrastructure that ships — document pipelines, LLM agents, data platforms, and full-stack product surfaces.
Client work · research · 10K+ daily requests in production
Featured Projects
Production systems · measurable outcomes · 2024—2026
COVID-19 Risk Profiles
Clustering 30M+ open health records to surface ICU risk patterns.
ML pipeline · Analysis · Dashboard · Next.js
NYC Ride-Hailing Analytics
Operational dashboards and ML models for Uber/Lyft trip and fare patterns.
Data science · Streamlit · ML models · Geospatial viz
India Air Quality Intelligence
Cloud ETL consolidating 500+ sensor streams into policy-ready indicators.
Data engineering · PySpark · Cloud architecture · Analytics
How I ship systems
Four layers I combine on every engagement — hover or tap to explore each one.
Machine Learning Pipelines
Training, evaluation, and production deployment with monitoring.
Built in EyeNet · COVID Risk Profiles
End-to-end orchestration: from EEG signal processing and feature engineering to production-ready LLM clusters and model deployment.
Data Engineering & Infrastructure
ETL, orchestration, and infra that survives real load.
Built in EyeNet · India Air Quality
ETL pipelines with PySpark and SQL, automated with n8n and Grafana, backed by Redis and high-performance GPU clusters.
Analytics Products
Dashboards and models stakeholders can actually use.
Built in NYC Ride-Hailing · COVID Risk Profiles
High-accuracy decision systems and predictive dashboards built with FastAPI, Streamlit and Power BI for real-world impact.
Scalable Web Platforms
Full-stack interfaces wired to AI and data backends.
Built in EyeNet · COVID Dashboard
Modern full-stack ecosystems using Next.js, React, and TypeScript, optimized for high performance and seamless AI integrations.
Hover to explore
How I think
Three rules I apply on every production system — after the demos are over.
“”
Principle 01
Ship the pipeline, not the notebook
Notebooks prototype. Pipelines with monitoring, alerts, and rollback are products.
A Jupyter notebook is a prototype. A pipeline with monitoring, alerting, and a rollback plan is a product. I optimize for the person who gets paged at 3 AM, not the one clapping at the demo.
Principle 02
Infrastructure is a feature
Deploy speed, observability, and GPU utilization are first-class engineering work.
Fast models mean nothing if your deployment takes 45 minutes and your GPU cluster idles at 12%. I treat provisioning, orchestration, and cost control as first-class engineering problems.
Principle 03
Explain it to the CEO or it didn't happen
If stakeholders can't understand the output, the model is broken — regardless of accuracy.
If a stakeholder can't understand why the model made that decision, the model is broken — regardless of its accuracy. Clarity is not documentation. Clarity is design.
Hover to read more
Practical systems, measurable outcomes, production from day one.
Tools in production
Every tool maps to real portfolio work — hover or tap to see where it shows up.
Hover a tool to see projects
Pick a tool
Select any tool from the bands above to see the featured projects that use it.
Let's work together
Open to AI engineering, data platforms, and full-stack product builds.
Freelance · contract · full-time