pprit.me/personal workspace
prit@portfolio: ~
system online
overview.sh
read, explore, connect.
prit@portfolio:~$ whoami
HELLO, WORLD. I'M

Prit
Panchal AI Engineer / Builder

Building LLM-powered automation,
RAG pipelines, and Python backends.

Generative AI · Multi-agent systems · Python

Vadodara, India·Building with purpose.
system_info.sh

❯ neofetch

██████╗ ██████╗ 
██╔══██╗██╔══██╗
██████╔╝██████╔╝
██╔═══╝ ██╔═══╝ 
██║     ██║     
╚═╝     ╚═╝     
prit@portfolio──────────────────

Role AI Engineer

Focus Generative AI

Lang Python

Shell Curiosity

Editor VS Code

Location India 🇮🇳

# currently building

◆ LLM-powered automation

◆ Reliable Python backends

❯always learning, always shipping.
01❯ cat about.md

Engineering intelligence.

A little context behind the commands.

I'm Prit, an AI Engineer who enjoys taking an idea from “what if” to a working system. I build LLM-powered automation and Python backends, with a focus on making the pieces work together.

At Navaera Software Services, I work on compliance-focused AI: combining deterministic transaction analysis with structured LLM narration. My approach is simple — let code handle the numbers, and give the model grounded evidence to work with.

Outside that work, I'm building Casey, a multi-agent task automation system with coordinated tools and sandbox execution.

Ground the output. Isolate the execution.
Build something that actually works.
impact.yaml INTERNAL TESTING
~5min

STR narration time
versus hours previously

5,000

transactions supported
per compliance case

14

rule-based detectors
deterministic Python logic

Observed in an internally tested AML/FINTRAC narration pipeline.

02❯ ls -la projects/

Selected AI & ML projects.

From coordinated agents to grounded answers. Things I've built.

4 directories
03❯ git log --oneline

AI engineering experience.

Learning by building, growing through real-world problems.

APR 2026 — PRESENTHEADa1e2026

Emerging System Engineer

AI Engineer

Navaera Software Services

  • Designed an AML/FINTRAC STR narration pipeline in internal testing, with observed narration time of approximately 5 minutes and support for up to 5,000 transactions per case.
  • Built 14 deterministic transaction-pattern detectors for structuring, layering, velocity anomalies, and FX discrepancies.
  • Separated calculations and evidence extraction from LLM narration, integrating locally hosted Qwen3-14B via vLLM with Pydantic-validated JSON and guided decoding.
PythonQwen3-14BvLLMPydantic
JAN 2026 — APR 2026b7f2026

AI Engineer Intern

Navaera Software Services

  • Contributed to compliance-focused AI engineering initiatives and transitioned into a full-time engineering role.
AI EngineeringCompliance
JUN 2025 — JUL 2025c3d2025

Machine Learning & Power BI Intern

Edunet Foundation · SAP-supported

  • Developed an interactive HR analytics dashboard with attrition, salary, and workforce KPIs in Power BI.
  • Built a used-car price prediction app using Random Forest, FastAPI, and a React/Tailwind frontend.
scikit-learnPower BIFastAPIReact
EDUCATION / 2022–2026

B.E. Information Technology

SVIT, Vasad

8.46/10CGPA
04❯ cat stack.json

Technical skills & AI stack.

The technologies I use to turn ideas into systems.

01

Languages

"languages": [
PythonSQLJavaScript
]
02

Generative AI

"generative_ai": [
LangChainLangGraphRAGQdrantPrompt engineeringStructured outputPydanticMCP concepts
]
03

LLM infrastructure

"infrastructure": [
vLLMQwenHugging FaceOllamaLM StudioDocker
]
04

ML & backend

"ml_backend": [
PyTorchscikit-learnspaCyFastAPIREST APIsGitReact.jsPower BI
]
✓ Always adding to the stack. Curiosity is a dependency.
05❯ ./contact.sh

Let's build something.

Have an interesting problem, a project, or an idea? Let's talk.

pprit2504@gmail.com
Say hello
# a good conversation starts here

def start_something():

idea = "your next big thing"

return collaborate(idea)

❯

prit@portfolio: ~

Welcome to my little corner of the internet.

Type help to get started. This shell explores the portfolio.

↑ ↓ historytab completeesc close