Parampreet SinghParampreet Singh

I'm the
AI Engineer

You're looking for :)

Building the bridge between humans and AI. I help people build their dream AI applications, be it - classical, Agentic or fine-tuned workflows with End-to-End deployment.

About

Portrait of Parampreet Singh

Parampreet Singh

Based in India

Resume

I am an AI/ML Engineer specialized in building production-grade AI systems. Currently, I'm focused on finetuning Small Language Models (SLMs) and scaling Gurmat Darbar, a full-stack platform for the Sikh community.
Educator at heart, I've delivered 70+ live sessions on Python and ML to 100K+ learners.
My work sits at the intersection of deep learning applications & scalable backend architecture.

BS in Data Science and Applications, IIT Madras (2022-Present)

LLMs, Mathematical Foundations for GenAI, and Software Engineering.

PyTorchLangGraphFastAPISnowflakeGCPFull-Stack Dev

Gurmat Darbar

Founder

Building the biggest digital ecosystem for the Sikh community.
Lead & engineered a Full-Stack platform to solve event discovery and data fragmentation using modern AI pipelines.

500+ Users in 1st Month

Since launch.

Largest Sikhi Samagam Platform

Covered 150+ Samagams with 200+ contributions.

Biggest Curated Sikhi Database

Structured multilingual data extraction.

FastAPIPostgreSQLRedisGCP Cloud RunGenAI Pipeline (EN/HI/PA)

Architecture focused on multilingual data extraction and auto-poster generation.

Gurmat Darbar logoGurmat Darbar platform preview

Work Experience

Founder

Gurmat Darbar

Sept 2025 – Present

Built a Full-Stack platform for the Sikh community featuring 'Find Samagams Near Me', covering 150+ Samagams with 200+ contributions.

Engineered a GenAI pipeline for structured multilingual data extraction and auto-poster generation.

AI Research Intern

mPragati Lab, IIT Delhi

Sept 2025 – Present

Working on 3D skull implant generation using deep learning for biomedical reconstruction.

Creating and training 3D U-Net based architectures for skull implant reconstruction.

Junior Applications Engineer

Coridors

Aug 2024 – Sep 2025

Built a Streamlit-based data ingestion platform with 30+ interactive screens integrated with Snowflake.

Worked on backend integrations with Snowflake Snowpark for scalable data workflows.

Designed, deployed, and maintained the company website using JavaScript and Tailwind CSS via Vercel and Cloudflare.

Educator

YouTube (@Param3021)

Jan 2023 – Present

Delivered 70+ live sessions on Python and ML, reaching 100K+ learners.

Teaching Impact

Python, machine learning, revision marathons, and the kind of teaching that turns exam fear into momentum.

YouTube

Parampreet Singh

@Param3021

Subscribers: 3.92k

Covers full Python - Basic to Advanced, full Machine Learning, Revision sessions, Course guidance, project guidance videos.

175K+

Learners

Reach

Concept-first teaching that helps learners move from basics to confidence.

70+

Live Sessions

Live

Regular live sessions covering Python, Machine Learning, and revision support.

500+

Student Feedbacks

Love

Real learner appreciation from doubt-solving, project guidance, and mentorship.

Wall of Love

Projects

Pocket-Coder preview

Pocket-Coder

A lightning-fast, 1.2B parameter local AI coding assistant designed to run flawlessly on CPUs and GPUs. Fine-tuned on a custom code-instruction dataset, it acts as a fully offline, zero-latency copilot inside Jupyter Notebooks and VS Code, completely bypassing cloud APIs.

Jupyter %%code Magic Command

VS Code MCP Server Integration

FinetuningSFTPEFT LoRALFM 2.5OllamaMCP
GRWM - Get README With Me preview

GRWM - Get README With Me

getreadmewithme.vercel.app

A production-grade, multi-agent GenAI platform that autonomously generates highly personalized GitHub READMEs. Orchestrated using LangGraph, it features three specialized AI agents working in tandem. The system utilizes real-time Server-Sent Events (SSE) for fluid streaming and is deployed securely on GCP Cloud Run.

LangGraph Agent Orchestration

100+ Unique Users in 16 Hours

LangGraphFastAPINext.jsGCPSSE
Hand Gesture LNN preview

Hand Gesture LNN

An exploratory deep learning research project implementing Liquid Neural Networks (LNNs) to handle 32-dimensional continuous-time motion data. By building custom Liquid Time-Constant (LTC) cells from scratch, the architecture improved temporal adaptability and significantly outperformed standard LSTMs.

Custom Liquid Time-Constant Cells

+9.6% Accuracy vs standard LSTM

RNNsLSTMLiquid Neural Networks (LNN)Time-Series Data
Quizzo-V2 preview

Quizzo-V2

A highly scalable, full-stack quiz management platform engineered with a robust asynchronous backend. It utilizes Redis and Celery for distributed background task queues to handle concurrent user loads, automating complex workflows like API rate limiting and dynamic certificate generation.

Redis & Celery Task Queues

Automated Email & Certificates

FlaskVue.jsRedisCeleryPostgreSQL