Vijay Kumar
I build AI systems and data platforms that are still running long after the demo ends.
Twenty years across full-stack, cloud data engineering, and now AI/GenAI — most recently building RAG pipelines and geospatial dashboards for aviation data at Shell. I write about where this is all headed, too.
Four things I actually know well
Not a skills cloud — the areas I've shipped production work in, repeatedly, under real constraints.
AI & RAG Systems
Retrieval-augmented pipelines, evaluation, and agentic workflows built to hold up outside a demo.
Cloud & Data Platforms
Azure Databricks and ADF pipelines, ETL architecture designed for scale, not for slides.
Application Engineering
React dashboards and geospatial visualization on data that genuinely needs to move fast.
Solutions Architecture
Turning fragmented systems into something one team can actually own and operate.
AeroCarbonIQ
A RAG-based decision support system for aviation fuel tankering and carbon strategy — built as a full-stack showcase of AI, aviation domain knowledge, and cloud data engineering.
Fuel tankering & carbon strategy, decision-support grade
RAG · Aviation data · Cloud-native
Pulls live aviation, fuel-price, and flight-tracking data into a retrieval-augmented pipeline so tankering and carbon-strategy decisions are grounded in current numbers, not stale spreadsheets. Built with a three-stage maturity model — local, serverless cloud, self-hosted GPU — so it can grow from a laptop prototype to a production deployment without a rewrite.
Latest from LinkedIn
Articles and "AI vs. Jobs" field notes, newest first — capped automatically at the 8 most recent.
Where this was built
Twenty years, four stops, one thread: shipping systems that keep working after launch day.
Senior Software Engineer
Azure Databricks and ADF pipelines, RAG-based GenAI tooling, and React dashboards with geospatial visualization for aviation fuel and carbon data.
Engineer
Retail AI systems recognized with a 2018 shop.org Innovation Lab award.
Engineer
Unity/C# game development, with a contribution credited on a filed patent.
Engineering Lead
Scaled a fantasy sports platform to 100k+ monthly active users; led carbon-aware compute scheduling that cut the platform's compute footprint, and DevOps changes that saved roughly £92k/year.