Lakshith Nishshanke
Engineer at Zeya Health
North Western Province, Sri Lanka
Summary
🤩
Rockstar🎓
Top SchoolLakshith Nishshanke is an engineer with eight years of practical software experience and a growing focus on AI and deep learning. He has shipped full-stack and iOS features at labml.ai, contributed a GPT-2 implementation and model refactors to a well-known annotated deep learning repository, and worked on algorithmic and IoT projects as a freelancer. Currently based in Sri Lanka and active at Zeya Health, he blends hands-on model experimentation with product-oriented engineering. Comfortable across ML research code, backend systems, and mobile, he has a knack for simplifying architectures—evident in his experiments replacing convolutions with linear layers and streamlining activation/dropout behavior.
7 years of coding experience
3 years of employment as a software developer
Bachelor's degree, Computer Science, Bachelor's degree, Computer Science at University of ColomboHigh School Diploma, Mathematics, 12, High School Diploma, Mathematics, 12 at Maliyadeva College
Stackoverflow
126reputation
14kreached
4answers
5questions
Github Skills (16)
attention-mechanism10
neural-network10
transformers10
pytorch10
deeplearning-ai10
gpt10
deep-learning10
artificial-neural-networks10
machine-learning9
ios6
winforms6
swift6
visual-studio6
multidimensional-arrays6
swiftui6
Programming languages (7)
Github contributions (5)
labmlai/annotated_deep_learning_paper_implementations
Jul 2024 - Aug 2024
🧑🏫 60+ Implementations/tutorials of deep learning papers with side-by-side notes 📝; including transformers (original, xl, switch, feedback, vit, ...), optimizers (adam, adabelief, sophia, ...), gans(cyclegan, stylegan2, ...), 🎮 reinforcement learning (ppo, dqn), capsnet, distillation, ... 🧠
Role in this project:
ML Engineer
ML EngineerContributions:1 review, 5 PRs, 1 comment in 29 days
Contributions summary:Lakshith implemented a GPT-2 model, likely for experimentation and demonstration purposes related to deep learning. Their contributions included creating a GPT-2 model implementation, loading pre-trained weights, and removing dropout layers and custom GELU implementations, indicating a focus on model architecture and optimization. The user also refactored the code, converting the convolutional layers into linear layers, and created an experiment notebook to test their changes.
notesdeep-learning-papersadamreinforcementswitch
lakshith-403/lakshith-403
Jan 2021 - Feb 2025
Contributions:12 pushes, 1 branch in 4 years 1 month
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.