Training new models

Hi, I'm Prathik.
An AI developer & DevRel.

Building production ML pipelines, fine-tuning LLMs, and deploying deep learning models at scale. I bridge the gap between cutting-edge ML research and real-world engineering.

Experience

Professional journey
and growth

A timeline of my professional journey through deep learning research, NLP engineering, and building production ML systems at scale.

ActualOne logo

ML Research Lead

ActualOne
Feb 2025 - Present

Leading LLM research initiatives at ActualOne, training transformer models on domain-specific corpora and building retrieval-augmented generation pipelines for enterprise AI products.

LLM TrainingTransformersRAG PipelinesDistributed Training
GenosisX logo

Co-Founder & ML Lead

GenosisX
Aug 2022 - Present

Co-founded GenosisX, building ML-powered products with a focus on model serving infrastructure, real-time inference APIs, and scalable feature stores for production ML systems.

MLOpsModel ServingFeature EngineeringSystem DesignEntrepreneurship
RisingWave logo

ML Engineer

RisingWave
Jun 2024 - Feb 2025

Built real-time ML inference pipelines and streaming feature stores using RisingWave's streaming SQL platform. Developed end-to-end demos for fraud detection and recommendation systems.

Streaming MLFeature StoresReal-time InferenceSQLData Engineering
Projects

Research & Engineering
Milestones

Each project represents a deep dive into ML research and engineering.
From building transformers from scratch to deploying models at scale.

View all projects
Transformer From Scratch: Implementing Attention Is All You Need
2024

Transformer From Scratch: Implementing Attention Is All You Need

LLM Fine-Tuner: LoRA/QLoRA Toolkit for Open-Source LLMs
2024

LLM Fine-Tuner: LoRA/QLoRA Toolkit for Open-Source LLMs

ML Pipeline Toolkit: End-to-End MLOps Platform
2024

ML Pipeline Toolkit: End-to-End MLOps Platform

Content & Speaking

Sharing knowledge
and insights

Talks, blogs, videos, and workshops where I share insights on deep learning, LLM training, and building production ML systems.

Scaling LLM Training to 1000 GPUs
talk
Dec 2024

Scaling LLM Training to 1000 GPUs

Deep dive into distributed training strategies, data parallelism, and pipeline parallelism for training large language models at scale.

LLMsDistributed Training+1
Understanding Attention Mechanisms: A Visual Guide
blog
Nov 2024

Understanding Attention Mechanisms: A Visual Guide

An illustrated walkthrough of self-attention, multi-head attention, and cross-attention with interactive visualizations and PyTorch code.

TransformersAttention+2
Hands-on LLM Fine-tuning with Hugging Face
workshop
Oct 2024

Hands-on LLM Fine-tuning with Hugging Face

A practical workshop on fine-tuning open-source LLMs using LoRA, QLoRA, and the Hugging Face ecosystem for domain-specific tasks.

Hugging FaceFine-tuning+2
Building Production ML Pipelines with Kubeflow
video
Sep 2024

Building Production ML Pipelines with Kubeflow

End-to-end tutorial on building automated ML pipelines with Kubeflow, from data ingestion to model serving with canary deployments.

KubeflowMLOps+1
Contact

Let's connect and
discuss opportunities

Open to research collaborations, consulting on ML infrastructure, and conversations about deep learning and MLOps.

Book a Call

Schedule a 30-minute call to discuss your project or potential collaboration.

Schedule on Calendly

Send an Email

Prefer email? Send me a message and I'll get back to you soon.

contact@prathikshetty.com
Send Email

Also available on social platforms