Building AI
that ships.
Not demos — production. I wire LLMs into real infrastructure: RAG pipelines, agentic workflows, vector stores, and the data that feeds them. Two and a half years of it at Factly. Now also building SadhanAI as a solo founder.
Tools I
trust.
AI / LLM
Backend
Data / Storage
Cloud / DevOps
Frontend
Auth
What I've
done.
- Built a reusable internal Pydantic validation library used across multiple data pipelines for schema enforcement, data quality checks, and error handling.
- Managed vector storage and semantic search optimisation using Qdrant and pgvector across multiple production AI applications.
- Developed a CRUD-based internal tool with FastAPI enabling 10+ non-technical team members to view PDFs and manage records without engineering support.
- Managed multiple simultaneous data pipelines with independent scraping schedules, validation rules, and storage targets — ensuring data quality using schema validation.
Work that
runs.
Click any project to open a full case study — architecture, workflow, screenshots, and the technical details.
SadhanAI
↗AI-powered coding-practice platform. Generates problems, test cases, and assignments from a single prompt.
Public-sector persona RAG
↗Agentic RAG over government PDFs — scrape → clean → embed → serve as a stateful chatbot.
WhatsApp scam-detection bot
↗Real-time conversational AI over scam-data PDFs, served on WhatsApp. Qdrant for semantic search.
Claude Code agents
↗Custom skills.md agents for code review, log triage, auto-PR creation. Cut manual eng time significantly.
Argo self-serve scheduler
↗Library that lets anyone schedule k8s workflows from a YAML config — zero Kubernetes knowledge needed.
Government PDF ingestion
↗ETL pipelines ingesting PDFs spanning 2000–2026. Weekly + monthly cycles, multi-store output.
Where it
started.
- Completed a specialisation in Data Science covering machine learning, data pipelines, and hands-on AI project work alongside core ECE degree.
- Electives focused on Artificial Intelligence, Machine Learning, and Python programming — which sparked my interest in AI and Data Science.
- Active participant in IEEE events and technical societies.
Bike
rides.
Long trips, solo. Helmet on, no itinerary.