Ashwin Mathur
Open Source Developer
Pune, Maharashtra, India
Summary
🤩
Rockstar🎓
Top SchoolAshwin Mathur is an AI engineer and open source developer based in Pune with four years of experience building and fine-tuning LLM-driven applications and RAG pipelines. He has contributed to high-profile projects like scikit-learn and deepset’s Haystack, improving documentation, backend serialization, and adding production-ready components such as Pinecone support and llama.cpp integration. His open-source research spans LLM fine-tuning (QLoRA, GRPO, DPO/PPO variants) and large-scale evaluation work—co-authoring MMTEB to benchmark multilingual embeddings across 1000+ languages and 500+ tasks. Ashwin combines rigorous academic performance (top of his MSc Data Science class) with practical production experience from internships and contributor roles, especially in retrieval, embedding selection, and evaluation metrics. He often focuses on infrastructure and evaluation hygiene—improving reproducibility, prompt optimization, and LLM-as-a-judge metrics—to reduce hallucinations and increase factuality in generative systems. Open to collaboration, he blends deep research experiments with pragmatic engineering to ship reliable LLM tooling.
4 years of coding experience
Master of Science - MSc, Data Science, 9.8 CGPA (Ranked 1ˢᵗ in Batch), Master of Science - MSc, Data Science, 9.8 CGPA (Ranked 1ˢᵗ in Batch) at Fergusson College10 CGPA, 10 CGPA at Delhi Public School, Pune
Github Skills (16)
scikit-learn10
machine-learning10
refactoring10
api10
pytest10
python10
apidoc10
documentation10
scikit10
nlp8
retrieval-augmented-generation8
statistics8
data-analysis8
data-science8
large-language-models8
Programming languages (3)
Github contributions (5)
deepset-ai/haystack
Jul 2023 - Apr 2024
AI orchestration framework to build customizable, production-ready LLM applications. Connect components (models, vector DBs, file converters) to pipelines or agents that can interact with your data. With advanced retrieval methods, it's best suited for building RAG, question answering, semantic search or conversational agent chatbots.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:22 reviews, 14 PRs, 55 comments in 8 months
Contributions summary:Ashwin primarily focused on refactoring code related to the serialization of API keys, specifically within the `AzureOCRDocumentConverter` and `SerperDevWebSearch` components. They removed API keys from serialization, renamed environment variables, and added tests to ensure proper initialization without API keys. Additionally, the user migrated the `RemoteWhisperTranscriber` to the OpenAI SDK, adapting the component to take ByteStream inputs and added a new `MarkdownToTextDocument` component. This work primarily involves improving the existing functionality and adding new features to the backend of the project.
pythonbertfinetuningproduction-readyhuggingface-transformers
scikit-learn/scikit-learn
Oct 2022 - Jan 2023
scikit-learn: machine learning in Python
Role in this project:Technical Writer
Technical WriterContributions:21 reviews, 12 commits, 14 PRs in 3 months
Contributions summary:Ashwin's commits primarily focused on improving documentation within the scikit-learn repository. They addressed documentation errors by ensuring that various functions and methods passed numpydoc validation. Furthermore, they updated user guide formulas, removed Matplotlib deprecation warnings from example code, and added examples and clarifications on Generalized Linear Models (GLM) topics. Their contributions highlight a focus on enhancing the clarity, accuracy, and completeness of the documentation for the library.
data-analysispythonstatisticsdata-sciencelearn-machine-learning
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.