Lead ML Engineer · Author · Mentor

Joydeep
Bhattacharjee

Lead ML Engineer @ Adobe — building intelligent systems at scale.
Author of 2 published ML books. Bengaluru, India.

2 Published Books
670+ Medium Followers
94 GitHub Repos
13+ Years in Tech
Book a Call Download Resume
Joydeep Bhattacharjee

About

Engineer, author,
and lifelong learner.

I'm a Lead ML Engineer at Adobe with over 13 years in the industry, building systems where machine learning meets real-world scale. My career spans TCS, HackerEarth, Nineleaps, Yellow.ai, and now Adobe — each chapter sharpening my focus on where systems meet intelligence.


I've authored two published books — Practical Machine Learning with Rust (Apress) and fastText Quick Start Guide (Packt) — and I write regularly on Medium about LLMs, ML system design, and the math behind modern deep learning.


Currently focused on LLM inference optimization, Flash Attention, quantization, and large-scale NLP pipelines. Available for 1:1 mentorship on Topmate.

Current Focus

LLM Inference Flash Attention Quantization NLP Pipelines ML System Design Diffusion Models Rust + ML

Currently at

Adobe

Lead ML Engineer

⬤ Active

Quick Stats

GitHub Stars 133+
Published Books 2
Medium Followers 670+
Open Source Repos 94

Writing

Recent Articles

March 2026

Understanding Flash Attention

A deep dive into Flash Attention — the algorithm that made training large transformers feasible by rethinking how attention is computed in memory.

Read on Medium
March 2026

Vera Rubin and the Future of AI

Exploring what the Vera Rubin Observatory teaches us about scale, data, and the infrastructure choices that will define the next generation of AI systems.

Read on Medium
September 2025

The Math Behind Diffusion Models — DDPM

Breaking down Denoising Diffusion Probabilistic Models from first principles — the forward process, reverse process, and the loss function that ties it together.

Read on Medium
August 2025

LLM Quantization Techniques

A practical overview of quantization strategies for large language models — INT8, GPTQ, AWQ — and how each affects inference speed and model quality in production.

Read on Medium
July 2025

Design: Harmful Ad Detection System

ML system design walkthrough for a scalable harmful ad detection pipeline — covering data flows, model selection, and trade-offs at production scale.

Read on Medium
June 2025

Design: Comment Ranking System

End-to-end ML system design for ranking comments at scale — feature engineering, model architecture, online/offline evaluation, and serving infrastructure.

Read on Medium
2024

LLM Quantization Explained

Shrinking AI models from feast to fit — a clear explanation of quantization fundamentals, why it matters, and how it reduces memory and inference cost without sacrificing accuracy.

Read on Medium
Read all posts on Medium

YouTube

Videos

Flash Attention
2:32:57

Deriving Flash Attention: The Math, the Hardware, and the Triton Implementation

75 views11 days ago
Diffusion Models
51:43

The MATH behind Diffusion Models DDPM

334 views5 months ago
LLM Quantization
59:04

LLM Quantization Techniques Explained — GPTQ AWQ GGUF HQQ BitNet

619 views10 months ago
LLM Quantization Explained
31:23

LLM Quantization Explained

448 views1 year ago
DeepSeek R1
18:12

The AI Breakthrough Nobody Expected: Inside DeepSeek R1

240 views1 year ago
Finetune Llama 3.2
35:40

Easily Finetune Llama 3.2 for Your Use Case | Upload to Ollama | Unsloth

1.7k views1 year ago
Multimodal AI
24:02

Meta's LLaMA 3.2: Big Leap into Multimodal AI | RUN LOCALLY

120 views1 year ago
GraphRAG
23:14

GraphRAG from Microsoft. Is it good? Implementation with Ollama + Llama 3.1

510 views1 year ago
CrewAI
16:34

MultiAgent AI System using CrewAI + Llama 3.1

475 views1 year ago
Watch all 85 videos on YouTube

Published Work

Books

Practical Machine Learning with Rust

Practical Machine Learning with Rust

83 stars 27 forks Apress

My first book bridges two worlds I love: Rust's performance guarantees and the richness of the ML ecosystem. It covers supervised, unsupervised, and reinforcement learning, computer vision, NLP, and deployment — all in Rust. Written for engineers who want memory-safe, high-performance ML beyond Python.

Springer Code Amazon
fastText Quick Start Guide

fastText Quick Start Guide

50 stars 10 forks Packt

A practical guide to Facebook's fastText — the library that made fast, scalable text classification accessible to everyone. Covers word representations, text classification, integration with Keras, TensorFlow, and PyTorch, and mobile deployment. A great entry point for applied NLP.

Packt Code Amazon

Open Source

GitHub Projects

practical-machine-learning-w-rust

Source code and examples for Practical Machine Learning with Rust — ML algorithms, computer vision, NLP, and deployment, all in Rust.

83 27
View on GitHub

fastText-Quick-Start-Guide

Code repository for the fastText Quick Start Guide — hands-on examples for text classification, word embeddings, and ML framework integrations.

50 10
View on GitHub

94 repositories and counting

View all repos

Career

Professional Experience

2022 → Present

Adobe

Lead ML Engineer

⬤ Current

Building and scaling ML systems at Adobe — focused on LLM inference, NLP pipelines, and intelligent content understanding. Research-to-production work with emphasis on performance, reliability, and scale.

LLM Inference NLP PyTorch ML Systems Hugging Face

Jan 2021 → 2022

Yellow.ai

Engineering Manager — NLP

Led the NLP infrastructure team building and deploying thousands of conversational AI models at scale. Architected multi-lingual, multi-modal NLP pipelines from research through Kubernetes-based production serving.

1000s of models at scale Multilingual NLP Kubernetes MLOps

May 2017 → Dec 2020

Nineleaps

Team Lead — ML Platform

Grew from individual contributor to technical lead on the ML platform — building production data pipelines, ML models, and leading the team through architecture reviews and mentorship.

ML Pipelines TensorFlow scikit-learn Team Lead

Nov 2016 → May 2017

HackerEarth

Category Head — Python

Built the Python practice section for B2C developer engagement, creating content and tooling for a high-engagement coding education platform.

Python EdTech

2011 → 2016

TCS & SLK

Systems Engineer

Early career in systems engineering at Tata Consultancy Services and SLK Software — building automation tools, configuring J2EE applications, and developing foundational software engineering instincts.

Java / J2EE Automation Systems

Expertise

Tech Stack

ML / DL Frameworks

PyTorch TensorFlow Hugging Face fastText scikit-learn

Languages

Python Rust SQL Bash

Infrastructure

Kubernetes Docker AWS GCP MLflow

Specializations

LLM Inference NLP Pipelines Quantization Flash Attention ML System Design

Mentorship

Let's work
together.

I offer 1:1 mentorship for engineers looking to grow in ML, navigate career transitions, or sharpen their system design skills. Whether you're preparing for senior interviews or shipping your first production model, I've been there.

ML system design reviews
Career guidance for ML engineers
Mock technical interviews
Resume and portfolio review
Book on Topmate

Book a 1:1 Session

Career guidance, technical deep-dives, interview prep, and more — on your schedule.

View availability on Topmate

Contact

Get in touch

Open to collaborations, speaking opportunities, and interesting conversations.

Email GitHub LinkedIn Medium X / Twitter Topmate