A distributed legal AI assistant for document query and generation
Concurrent document processing across Go and Python services, with retrieval-augmented generation over legal documents.
- Go + Python concurrent processing services
Natural-language analytics over IoT time-series data
An analytics platform that lets operators ask questions of equipment telemetry in plain English.
We do not publish client names
The problem
Equipment generated roughly ten thousand records a day, and getting an answer out of that data meant writing queries, which put analytics out of reach of the people who understood the equipment.
What we built
End-to-end ETL pipelines on Kafka for collection, validation and enrichment of IoT payloads. Time-series modelled in TimescaleDB with indexing and retention policies, and Druid datasources with rollups for fast exploratory analytics. FastAPI services expose the cleaned datasets and natural-language query endpoints; a language model translates plain-English questions into executable queries and summarises the results. The full interface was built in React and Vite.
The result
Operators query telemetry conversationally, with history and export flows. Services are containerised and orchestrated with Kubernetes, deliberately environment-agnostic rather than tied to one cloud.
More work
Workout360: gym and studio software with live classes, built for India
Our own product, designed and built end to end: QR check-in with no hardware, GST-compliant invoicing, WhatsApp reminders and live video classes inside the operator's own app.
- 1 sec to check a member in
- Rs 0 of check-in hardware required
A real-time regulatory data platform for compliance teams
Distributed Go services processing regulatory rules for search, sharing and management on one platform.
- Real-time regulatory data streaming
Tell us what you are trying to build
Describe the system and the constraint you have hit. You will get a technical reply, not a sales sequence.