About
The engineering around the model.
Not the LLM is my public notebook for the parts of AI systems that decide whether a promising demo becomes a useful product: context, tools, runtimes, evals, observability, governance, and the human workflows around them.
About Me
Shubham Jindal
I am Director of AI Software Engineering at Harness. I work on the systems around AI products: context, tools, agent runtimes, evaluation, observability, governance, and the platform architecture that makes models useful in production.
This site is where I turn that work into public notes: essays, diagrams, and practical frameworks for engineers building with models without treating the model as the whole product.
Why this exists
The model matters, but most of the hard engineering lives in the system around it: how context is retrieved, how tools are exposed, how agents are steered, how failures are observed, and how teams keep the whole thing understandable enough to operate.
This site is where I work through those questions in public, with a bias toward practical architecture over launch-day spectacle.
What I care about
- Model Context Protocol and tool contracts
- Agent loops, harnesses, and runtime design
- Evaluation, observability, and reliability
- Governance patterns engineers can actually use
- AI-native software delivery and platform architecture