Deep Learning Engineer

Prajwal Anagani

I build production-grade ML: data quality, evaluation, deployment, and iteration loops that actually ship. Proud open source contributor.

Get in touch Read the blog
Fortune 10 Clients
10+ Production Models
ACL & IEEE Published
HuggingFace
Contributor
PythonPyTorchCUDATritonvLLMC/C++RustDockerKubernetesAWSAzure
LawJarp-A's GitHub contribution chart

Experience

Deep Learning Engineer

RadiusAI · Jun 2024 – Present

  • Built 10–15 production ML models for Fortune 10 clients handling 5K SKUs with 90% precision
  • Process millions of images monthly via Triton with sub-1s latency
  • Reduced data corruption by 60% through automated quality pipelines
  • Accelerated training 1.5x and integrated VLMs for +3% accuracy

ML Developer

CARAIO Technologies · Jun 2022 – Aug 2024

  • Led team building chromosome analysis pipeline: 85% segmentation, 98% classification
  • Built prototype to shipped product within 1 year
  • Part of founding team at UC Berkeley SkyDeck incubated startup
  • End-to-end: metaphase images to karyotype generation

Open Source

View GitHub →
Featured

HuggingFace Diffusers

Added caching system accelerating inference 1.5–2.5x. GPU acceleration of legacy components.

PythonPyTorchDiffusion
View contributions →

vLLM

Contributing to vllm-omni: caching for DiT model inference, added FLUX 2 model support.

PythonCUDALLM
View contributions →

Ground-up Series

Reproducing foundation model architectures from scratch with clear, runnable notebooks.

PyTorchTransformers
View project →

Projects

Chromosome Classification

Medical CV classification with focus on accuracy under data constraints. Part of CARAIO's diagnostic pipeline.

PyTorchMedical CVClassification
View project →

MWP-T5

Fine-tuned T5 for generating solvable math word problems. Published at IEEE conference.

T5NLPFine-tuning
View project →

Publications

Chromosome Segmentation Analysis Using Image Processing Techniques and Autoencoders

Research Paper

Writing

Thoughts on deep learning, ML systems, and engineering.

Hello World (Yes, Really)

Why a Deep Learning Engineer who talks to machines all day decided to start talking to humans. Spoiler: the machines suggested it.

Oct 2024
View all posts →

Let's connect

I'm always interested in discussing deep learning, production ML, or potential collaborations.

Send an email GitHub LinkedIn