Stay curious with

Presenting,

KhushviBamrolia?

An AI software engineer who builds the hard parts from scratch — a query language in Rust, conversational and voice AI in production, and the whole engine room that keeps them running.

EST. 2024

VALUE — priceless

गुजरात, भारत

scroll · आगे

The Builder's Manifesto · घोषणापत्र

I don't just use frameworks. मैं build them.

Backend & Systems Engineer → building databases, query runtimes & AI infrastructure from scratch.

Born in Gujarat · forged in code

परिचय About / whoami № 01

I build the hard parts from scratch and own them end to end.

My work lives where AI meets systems engineering.I took a Industry AI product from no conversational capability to production systems from owning LLM selection, context management, dialogue state and low-latency deployment.

I'm drawn to foundations. CoffeeQL, my own query language, has a hand-written lexer, parser, planner and executor in Rust and federates across four databases under one syntax. When a problem needs a primitive that doesn't exist, I write it.

I care about the team around the code too I've introduced code-review standards, QA frameworks and Algorithm documentation processes from scratch, and mentored 100+ engineers through data structures and system design.

गिनती By the numbers № 02

1000/1000

tests passing

250K+

records searched

<200ms

retrieval latency

databases, one syntax

100+

students mentored

9.77

diploma CGPA

काम Chapters of Innovation № 03

♨ Coffee
QL query · lang

100% hand-written · no ORM

Chapter I · The Quest for One Syntax

CoffeeQL

one syntax, every database

A query language and orchestration engine in Rust — lexer, parser, planner, executor and adapter layer — unifying SQL, document and key-value stores under one chainable syntax.

user[] · products{} · ns:id{}

4 native adapters 500/500 tests compiled to WASM 500+ weekly npm

coffeeql.dev ↗ npm ↗ PyPI ↗ GitHub ↗

खोज <200ms

250K+ records · custom relevance

Chapter II · The Quest for Latency

Content Search Engine

fast where it counts

An AI-enhanced website search engine — scrape, index, structured retrieval — at sub-200ms latency. I designed stop-word filtering and autocomplete deduplication from scratch for measurably higher precision across 250K+ records.

<200ms retrieval 250K+ records custom relevance scoring

⚙ rust · pipe

concurrent · <100ms

Chapter III · Built for Throughput

Data Transformer

built for throughput

A Rust data-transformation system built on modular concurrency pipelines for distributed processing, holding sub-100ms latency under load.

<100ms under load concurrent pipelines Rust

काबिलियत What I can do № 04

Conversational & Symbolic AI

Take AI products from zero to live systems from LLM selection, context & dialogue-state management, and symbolic pipelines that turn messy sources into structured, auto-generated documents.

Query Languages & Runtimes

Design languages from the ground up from hand-written lexers, parsers, planners and executors and federate across SQL, document and key-value stores under one syntax.

iii

Search & Retrieval

Build fast, precise search across hundreds of thousands of records and custom relevance scoring, stop-word filtering and autocomplete dedup, tuned for sub-200ms retrieval.

High-Performance Systems in Rust

Concurrency pipelines and data-transformation systems that hold low latency under load, built for throughput, correctness and a clean compile.

Infrastructure & Reliability

Own CI/CD, deployment workflows and service-health monitoring end to end, keeping production systems running with zero downtime.

Engineering Culture

Set up code-review standards, QA frameworks and ADR processes from scratch, and mentor engineers and students through DSA, architecture and scalability.

Want the full story including roles, timelines, the whole résumé?

Let's connect on LinkedIn ↗ — message me and I'll share my résumé

हुनर The ingredients № 05

B.Tech — CSE

Darshan University

Diploma — CSE

Gujarat Technological University

9.77 CGPA

नमस्ते —

Let's build कुछ real.

bamroliakhushvi@gmail.com

GitHub ↗ LinkedIn ↗ coffeeql.dev ↗

Khushvi
· 2026 ·
India