Tanay Mehta

Berlin, Germany

Join Prog.AI to see contacts

Summary

🤩

Rockstar

🎓

Top School

Tanay Mehta is an experienced ML engineer based in Frankfurt with nine years building and shipping production-grade AI systems across startups and research-focused companies. He has hands-on experience scaling agentic chat systems, distributed LLM pre-training and on-demand fine-tuning, and improving RAG/document-chat pipelines for enterprise use. His open-source contributions include integrating PoolFormer into the Hugging Face Transformers library and adding a jittable hinge loss to DeepMind’s Optax, signaling deep familiarity with model internals and optimization. Past roles span practical retrieval and LLM deployments—building internal RAG search, code-search, and large-document tooling—and achieving strong retrieval metrics using memory‑efficient inference. He blends research-grade technical rigor from an MSc in Data Science with pragmatic product delivery, often bridging core model work and developer-facing APIs. A less obvious strength is his pattern of improving tooling and automation (fine-tuning APIs, file-format integrations) that multiplies team productivity beyond individual model improvements.

9 years of coding experience

3 years of employment as a software developer

Bachelor of Technology - BTech, Computer Science, Bachelor of Technology - BTech, Computer Science at JECRC University

Masters of Science (MSc), Data Science, Masters of Science (MSc), Data Science at University of Bath

English, Hindi

Github Skills (14)

pytorch10

machine-learning10

deeplearning-ai10

transformer10

nlp10

deep-learning10

jax10

python10

optimization10

natural-language-processing9

pre-trained-model9

numpy9

hub9

language-model9

Programming languages (11)

C++CSSRustCTeXJavaScriptPHPHTML

Github contributions (5)

google-deepmind/optax

Aug 2022 - Nov 2022

Optax is a gradient processing and optimization library for JAX.

Role in this project:

ML Engineer

Contributions:3 reviews, 8 commits, 3 PRs in 2 months

Contributions summary:Tanay primarily contributed to implementing and refining the hinge loss function for binary classification within the Optax library. Their work involved iteratively adding the function, making it jittable, writing tests, and fixing related issues. The contributions highlight an understanding of loss functions within the context of machine learning optimization, a core aspect of the repository's purpose. The user's work also touched on documentation and API integration.

deep-learningoptimizationmachine-learningoptaxjax

huggingface/transformers

Feb 2022 - Feb 2022

🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.

Role in this project:

ML Engineer

Contributions:26 reviews, 2 commits, 12 PRs in 8 days

Contributions summary:Tanay contributed significantly to the integration of the PoolFormer model into the Hugging Face Transformers library, implementing new features such as `PoolFormerFeatureExtractor` and integrating it with the AutoFeatureExtractor API. They addressed multiple bugs and resolved compatibility issues, fixed documentation, and incorporated the model into the library's structure. The user's work involved modifying core modeling files and testing components to ensure full functionality of the PoolFormer model. They have added support for the Fill-in-the-middle training objective and related code.

pythonbertspeech-recognitionstate-of-the-artflax

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.

Request Free Trial