Hi, I’m
Praneeth.
A little curiosity.
A lot of experiments.
AI Solutions Architect
Databricks · Bengaluru
I build AI systems.
Then I find out
what makes them work.
Experiments in agents, memory, and evaluation. Code, results, and the things I learn along the way.
Explore my writingWriting
Before the Answer: How a Language Model Is Made
One question travels from raw web text through tokens, attention, cluster-scale training, and feedback to its first streamed answer. Sixteen animated drawings, three things that break, and a scaling lab.
Databricks Is Databricks for Agents
Build and run agents with governed context, agent execution, MLflow, AIR with NVIDIA, and open-model serving. Explore the platform and how its capabilities connect.
Reading Tastes, Ranking Titles
How language models became recommenders, and what Netflix’s GenRec changes. Explore catalog scoring, training, cost, and the research lineage.
WTF Is a System One Model?
What TypeSafe means by System One, why anyone wanted a model that cannot write a sentence, and what happened when I put Jev and GPT-6 Luna in front of the same two routing jobs.
Adapting Jev to Your Domain with GEPA
A practical guide to Jev and domain adaptation with GEPA, from defining the task to measuring results on medical literature.