Undergrad
B.Tech in AI & ML
Learned that models are easy and production systems are hard — and which one I wanted to get good at.
About
I build the parts of software that have to be right.
Before the UI, before the demo, there is a schema, an API, and a decision about how data flows. That is where quality actually lives — and it's where I spend my time.
RoleAssociate SWE · Peoplebox.ai
BaseVijayawada, India · Remote
StackRails · React · SQL · RAG
Patents2 filed
The arc
From AI theory to production trust.
Jan 2025
SWE Intern · Peoplebox.ai
Learned enterprise SaaS on the job, shipping real customer-facing work instead of toy projects.
Jul 2025
Associate Software Engineer
Owned backend services, REST APIs, and AI features in production across Goals, Reviews, Surveys, and HRIS.
Now
Enterprise AI, in production
Two patents filed. RAG-powered workflows shipping. Optimizing for reliability — the unglamorous stuff.
How I think
Three beliefs, held tightly.
Data models are product decisions
A schema outlives the code that reads it. I design it like a contract — because it is one.
Reliability is the feature
The best system is the one nobody notices. My job is to make boring look like good luck.
AI is an engineering tool
LLMs are only useful grounded, evaluated, and measurable. RAG without retrieval quality is a guess generator.
Principles
The rules I write code by.
Clean, production-ready code
A feature that ships on time and stays reliable beats a clever one that needs a caretaker.
Data through the same lens
Well-designed schemas and optimized queries matter as much as the product surface they power.
AI grounded in reality
Retrieval, evaluation, and prompt design come before the model. Grounded beats impressive.
Reliability over cleverness
The best system is the one nobody has to think about because it just keeps working.
Optimizing for
What I actually measure.
Ship velocity90
Move fast enough that feedback is real; slow enough that production stays boring.
Code craftsmanship82
Write code the next engineer can read without a sigh.
Product thinking76
The metric, not the ticket, is the requirement.
AI grounded in data84
Retrieval and evaluation quality before prompt quality.
Reliability at scale88
Design for the load you hope to have — then make it boring.
Workflow
From idea to shipped.
Idea
Start from the user problem, not the feature.
Design
Schema and API contracts first. Interfaces are the product.
Build
Small, readable, testable increments. Review like a skeptic.
Measure
Profile against real load: query plans, latency, edge cases.
Ship
To production, then iterate on what users actually do.
Lessons
What production taught me.
The query plan is the truth
Optimizing SQL taught me the schema is worth more than clever code. The database doesn't lie.
Ground the model or don't ship it
RAG only works when retrieval quality is measured. I learned that the hard way — once.
Production issues are feature requests
Every incident taught me more about the domain than the ticket did. I read the logs first.
Refactor like you own it
Code you inherit is code you ship. Leave every file a little better than you found it.
Developer DNA
How I'm wired.
Backend systemsRails · REST · Sidekiq
Databases & SQLMySQL · PostgreSQL · indexing
AI / RAGLLMs · embeddings · evaluation
Frontend (React)interfaces that ship
Product sensemetrics over opinions
Curiosityreads the codebase, not the FAQ
Beyond the code
What shapes the thinking.
AI tools for my own workflow
Building small agentic helpers to see what breaks — and what holds.
How enterprise systems scale
Reading how real products grow their data, queues, and teams.
Distilling complexity
Writing and explaining until a hard problem sounds obvious.
Now · live status
Building enterprise AI features, learning advanced RAG and vector search, and exploring agentic AI tools.
Enter the studio →The best way to know how I work is to work together.
Let's talk →