Joydeep Bhattacharjee — Lead ML Engineer

Currently a Lead ML Engineer at Adobe, where I build core GenAI services and APIs powering Adobe Firefly. My day-to-day involves architecting enterprise-scale ML workflows for model customization and serving, and optimizing GPU-accelerated pipelines for training and inference using PyTorch, CUDA and Triton. I work on what happens behind PyTorch — the GPUs, kernels, memory hierarchies, and distributed systems that make AI actually work at scale. I’ve spent the last decade going deeper down the stack — from building NLP systems and demand forecasting models earlier in my career, to inference optimization on NVIDIA A100s, H100s and Blackwell and working at the intersection of generative AI and GPU systems engineering. Topics I cover regularly: GPU architecture (SMs, register files, memory hierarchies), distributed training infrastructure (NCCL, InfiniBand, 3FS), inference optimization (KV cache, quantization, FlashAttention), and the latest from NVIDIA, Google TPU, and DeepSeek. On Medium and youtube, I go through long-form technical deep-dives on topics like FlashAttention internals, diffusion model math, LLM quantization (GGUF, GPTQ, AWQ, BitNet), and ML system design. Author of two books: • fastText Quick Start Guide (Packt) • Practical Machine Learning with Rust (Apress) Find me here: 🌐 Website: https://infinite-joy.github.io/ 📝 Medium: https://joydeep31415.medium.com 🎥 YouTube: https://www.youtube.com/channel/UCkgbnb9ibABSL5X9Au-8mIA If you care about making AI faster and cheaper so that the defining technology of our generation reaches to more people, let’s connect.

Stackforce AI infers this person is a SaaS and Semiconductor expert with a focus on AI and ML optimization.

Location: Bengaluru, Karnataka, India

Experience: 14 yrs 4 mos

Skills

Artificial Intelligence (ai)
Machine Learning
Deep Learning
Natural Language Processing (nlp)
Statistical Data Analysis

Career Highlights

Expert in architecting enterprise-scale ML workflows.
Proven track record in optimizing GPU-accelerated pipelines.
Author of two influential machine learning books.

Work Experience

Adobe

Lead Machine Learning Engineer (9 mos)

Samsung Semiconductor India R&D

Senior Machine Learning Staff Engineer (3 yrs 2 mos)

yellow.ai

Engineering Manager (NLP) (1 yr 6 mos)

Nineleaps

Team Lead - Retail Demand Forecasting (1 yr 8 mos)

Team Lead - Models as a Service for Medical Analytics Application (2 yrs)

HackerEarth

Category Head - Python (6 mos)

SLK

Software Engineer (1 yr 2 mos)

Tata Consultancy Services

Systems Engineer (3 yrs 7 mos)

Education

Bachelor of Technology (B.Tech.) at National institute of Technology Silchar

Joydeep Bhattacharjee

Lead ML Engineer

Bengaluru, Karnataka, India14 yrs 4 mos experience

Key Highlights

Expert in architecting enterprise-scale ML workflows.
Proven track record in optimizing GPU-accelerated pipelines.
Author of two influential machine learning books.

Stackforce AI infers this person is a SaaS and Semiconductor expert with a focus on AI and ML optimization.

Contact

Skills

Core Skills

Artificial Intelligence (ai)Machine LearningDeep LearningNatural Language Processing (nlp)Statistical Data Analysis

Other Skills

PyTorchComputer VisionCUDATritonRetrieval-Augmented Generation (RAG)QuantizationGraph optimizationDemand ForecastingApache SparkPythonSparkAWSPython (Programming Language)performance optimisationDiffusion

About

Experience

14 yrs 4 mos

Total Experience

Average Tenure

Current Experience

Adobe

Lead Machine Learning Engineer

Sep 2025 – Present · 9 mos · Bengaluru, Karnataka, India · Hybrid

Lead technical development of core GenAI services and APIs integrating generative models on Adobe Firefly
Architecting and developing enterprise-scale ML workflows for model customization, serving, and
ecosystem integration with both Adobe’s first-party and third-party generative models
Building and optimizing GPU-accelerated pipelines for model training and inference with focus on
performance, scalability, and reliability using PyTorch, CUDA, Triton, and TensorRT
Providing hands-on technical leadership to engineering team, driving architecture decisions, design
reviews, and technical standards for high-reliability systems
Researching and evaluating emerging ML and MLOps technologies to enhance engineering velocity and
system performance across the organization
Driving cross-functional alignment with Product Managers, TPMs, and engineering leaders to define
and deliver on GenAI Services roadmap
Leading team in tackling complex engineering challenges related to diffusion models, transformers, and
optimizing inference latency and throughput at scale
Fostering culture of innovation and technical excellence while mentoring ML engineers in distributed
systems, Kubernetes, and GPU resource management

PyTorchComputer VisionArtificial Intelligence (AI)Machine Learning

Samsung semiconductor india r&d

Senior Machine Learning Staff Engineer

Jul 2022 – Sep 2025 · 3 yrs 2 mos · Bengaluru, Karnataka, India · Hybrid

Performance tuning and inference optimisation for state-of-the-art LLM models
for in-house NPU computer architecture using technologies such as quantization and graph optimization
Research and implementation of novel Retrieval Augmented Generation (RAG) + LLM application for
breakdown maintenance of semiconductor manufacturing equipment. Built from scratch
with precision of 96%. Applied LLM Quantization techniques on llama3 model to bring response time
from 30 seconds to less than 10 seconds by 70%.
Applied Research in the area of Deep Learning and its application in semiconductor manufacturing
processes. Responsible for improving FAB yield using Deep Learning and AI models. Work on original
research. As part of project performed more than 100 experiments in 2024.
Handle large datasets and develop data pipelines to provide inputs for training and testing models.
Lead and mentor a team of AI researchers, define technical roadmap and architecture for AI projects.
Prepare and submit presentations and project reports to upper management.
Collaborate with cross-functional teams, including product managers, and front-end developers, to
design solutions that meet business goals.

Deep LearningRetrieval-Augmented Generation (RAG)PyTorchNatural Language Processing (NLP)Artificial Intelligence (AI)Machine Learning

Yellow.ai

Engineering Manager (NLP)

Jan 2021 – Jul 2022 · 1 yr 6 mos · Bengaluru, Karnataka, India

Machine Learning ‐ Deep Learning ‐ Development and Team Mentoring
Responsible for research, development, production, and scaling of full pipeline of message flow and other intelligent NLP systems in multi‐lingual and multi‐modal contexts.
Research, implementation, benchmark existing literature for various purposes.
Leading the team which works in parallel and in collaboration to produce custom‐made algorithms using machine learning/Deep learning frameworks‐libraries like torch/hugging face/TensorFlow.
Prepare production‐ready code with pre/postprocessing and native scaling in Kubernetes or on‐prem environment.
Responsible for all MLOps, Cloud infra management for ML‐related service.
As a lead, I am answerable for client issues ﴾ explanation on models prediction and performance﴿, custom requirements, new product feature exploration, POCs, architecture and design discussion, problem solving, hiring, mentorship and other decision making tasks.

Natural Language Processing (NLP)Deep LearningMachine LearningArtificial Intelligence (AI)

Nineleaps

2 roles

Team Lead - Retail Demand Forecasting

May 2019 – Jan 2021 · 1 yr 8 mos · Bengaluru, Karnataka, India

Design ML and Optimization architectures for Inventory Forecasting and Optimization system for 40000 products across thousands of stores.
Design solutions for enhancements to the forecasting model.
Deployment and maintenance of highly available forecasting application.
Led POCs for exploring new forecasting methods and technologies.
Time Series analysis of various KPIs.
Assisting business with Forecast accuracy reporting.
Build a team from scratch for the successful delivery of the product.
Tech Stack: Python, Spark, Hadoop, Teradata, Hbase, Jupyter

Demand ForecastingApache SparkMachine LearningStatistical Data Analysis

Team Lead - Models as a Service for Medical Analytics Application

May 2017 – May 2019 · 2 yrs · Bengaluru, Karnataka, India

Architected and built model-serving infrastructure for NLP models
Auto-scaling cluster of 50+ deep-learning models
Large-scale online and offline serving meeting throughput of 1M+ prediction requests a day
Built the model deployment lifecycle from ground up: model versioning, deployment, monitoring and dashboarding
Enabled push button deployment of ML models
Public health knowledge graph:
Built ETL pipelines, data-access and querying layer for graph DB at a scale of 1B+ edges and 200M+ nodes
Tech Stack: Python, AWS, Spark, Pandas, Neo4j, Airflow

Natural Language Processing (NLP)Machine LearningArtificial Intelligence (AI)