Jagadeesh Rampam

The case for TypeScript SDKs in Grove

An evidence-led technical deep dive and a dogfooding exercise in building Grove's agent pipeline with its own research agent — Grove 0.2.0 | Finalised 7…

17 hrs ago • Parjanya by Phagyul AI

1

ಕೊಡಗಿನ ನಗು | Kodagu Smiles

Beyond the landscapes, a portrait of the people, everyday lives and quiet diversity of Kodagu.

Nov 6 • Parjanya by Phagyul AI

1

Art of Seeing

From mastering the technique to discovering the story

Nov 3 • Parjanya by Phagyul AI

1

Building Grove v0.1: The Agent Harness as a First-Class Citizen (as well as first principle)

From architectural principles to local experimentation with Gemma4:31B, Ollama, and an in-house research agent in the pipeline. More agents to follow…

Nov 2 • Parjanya by Phagyul AI

1

2

September 2026

How We Cut AI Inference Cold Starts in Half: Optimising GPU Startup for Photo Analysis

Two months of measuring, profiling and simplifying Parjanya’s AI infrastructure — from GPU cold starts and model loading to inference speed, cost, and…

Sep 30 • Parjanya by Phagyul AI

2

2

2

The Image Was Fine. Our Pipeline Wasn't.

How a 16-bit grayscale TIFF exposed a hidden failure in our preview pipeline—and why fixing one line of code wasn't enough.

Sep 24 • Parjanya by Phagyul AI

Parjanya 2.0 is open to everyone

Release notes from seven weeks of closed beta: 18,884 images, 18+ formats, 20+ photographers, and what they contributed.

Sep 14 • Parjanya by Phagyul AI

August 2026

Price Is Only Half the Compute Problem

Why the right infrastructure decision is not “cheapest instance,” but the right capacity at the right price and the right time.

Aug 24 • Parjanya by Phagyul AI

1

N+1 Without an ORM: The Hidden Cost of Extra Round Trips

How we found duplicate DynamoDB reads, unbatched CloudWatch calls, and a cache that quietly wasn’t there

Aug 18 • Parjanya by Phagyul AI

1

1

From Managed Services to Primitives (DX vs DIY)

What Building Parjanya v2.0 Taught Me About Managed Platforms, Self-Hosted AI, and Owning the Right Abstractions

Aug 17 • Parjanya by Phagyul AI

1

June 2026

When SageMaker Wasn't the Problem

The Parjanya v2.0 case study: how AWS primitives replaced a managed ML platform through reconciliation, replay, and event-driven orchestration

Jun 24 • Parjanya by Phagyul AI

1

From Pipeline to Platform

Event-Driven TBIE Architecture, Reconciliation Patterns, and the Lessons from Building Parjanya v2.0

Jun 22 • Parjanya by Phagyul AI

1