Companies
Long-term ambitions, with room to grow into something real.
The long viewSoftware × AI × Products
I’m Prathmesh Adsod, a software engineer and product builder. I work on AI agents, Python and Java backend systems, and the product experience around them. My focus: software that earns its place in someone’s workflow.
Long-term ambitions, with room to grow into something real.
The long viewIdeas in exploration. Useful things earn their public release.
What comes nextOriginal projects and experiments. The code tells the story.
Explore GitHubA home for the decisions, questions, and lessons along the way.
Notes & newsletterIn progress
I’m exploring where AI capability meets real engineering constraints—and where a thoughtful product can make the difference.
Memory, tools, evaluation, and clear boundaries. What helps an AI system do useful work you can actually inspect?
From an answer on a screen to a workflow that handles data, permissions, failures, and the next step.
Useful interfaces, honest states, and fewer unnecessary decisions. Complexity belongs in the engineering.
Selected public work
Can an agent learn from its past work? Can a decision be traced to evidence? Can a system stay inside its constraints? These projects put those questions into code.
All public repositoriesExecution memory for Agent Skills: scoped recall, run evidence, recovery paths, and invalidation for a query-regression workflow.
A dataset release-review workflow combining source context, measured checks, human approval, and hash-verifiable release evidence.
Off-grid solar design that normalizes web component inventory and checks panel, inverter, and battery combinations against electrical constraints.
How I contribute
Build the parts around the model: retrieval, tool access, memory, and evaluation. Make the behavior understandable enough to improve.
Connect APIs, data models, and durable workflows, with deliberate handling of failures and changing requirements.
Connect the backend to an interface people can navigate, including the loading, empty, and error states.
Use focused experiments to answer a specific question before committing to a larger build.
A principle I build by
“The same script can become a good film or a masterclass. The difference is execution. I build for the masterclass.”
The idea is a starting point. Architecture, judgment, simplicity, and thousands of small decisions determine what it becomes.
More about my approachThinking in public
Trade-offs, experiments, and the questions that change a project’s direction. Essays and engineering notes will live here as they’re published.
Writing & newsletterA little about me
I like the moment before a solution is obvious: asking better questions, testing assumptions, and deciding what deserves to be built. That curiosity connects my work in engineering, applied machine learning, and product design.
The longer introductionLet’s connect