About

Hi, I’m Aakash! I hold a B.Tech in Data Science and AI and work as an ML Engineer at Needl, where we are working to solve information overload problems in the finance domain. As part of my role, I work on developing and deploying NLP solutions, working on document AI, and more recently, around agents for automated insights generation.

During my bachelors, I have had a good time exploring Linear Algebra, Probability & Statistics, classical ML, Adversarial ML, GPU architecture (using CUDA) and Psychology.

I like reading crisp research papers, especially when I can quickly try out and apply their ideas and insights to real-world problems.

Interested in Collaborating?

📄 View my Resume here

Open source highlights

• November 2024
  • Added model2vec embedding support to chonkie semantic chunker.
• May 2024
  • Created a summarization metric to evaluate summarization tasks, now available in RAGAS.
  • Integrated epsilla retrieval module in dspy.
• February 2024
  • Came up with an idea of a new entity-based RAG evaluation metric and implemented in RAGAS. This metric is useful in RAG use-cases where entities are of importance and hence measuring their coverage by the context is important.
  • Expanded GPTCache with Marqo integration for vector store and Nomic integration for embeddings.
• January 2024:
  • Created a smart query routing method to route query to most relevant sink in Neum AI.
  • Implemented BLEU and METEOR metrics for machine translation in UpTrain AI.
  • Created Marqo vector store integration for EvaDB.
• December 2023
  • Expanded filtering in Neum AI by adding unified filter.
  • Integrated LanceDB sink and Marqo sink to Neum AI.
• November 2023
  • Integrated marqo tensor search as retrieval mechanism in dspy.
• May 2023
  • Developed approx-KNN based graph construction to pytorch-geometric.
• July 2022
  • Implemented new distance metrics to weaviate.
• May 2022
  • Developed an example application - an attendance system - using a vector DB. It was also used by weaviate’s GSoC contributor that year.
• March and April 2022
  • Developed a multi-modal search example which was also used in weaviate’s GSoC project.