Dyotak Kachare

Passionate about Deep Learning and Complex System Architectures.

Dyotak Kachare

About Me

Hi! I'm a passionate AI & ML Engineer with a keen eye for creating elegant, efficient solutions to complex problems. Currently pursuing my Bachelor's in Artificial Intelligence and Data Science with Honours in Blockchain at VESIT, Mumbai.

Languages

Python
Java
JavaScript
C

Frameworks

PyTorch
TensorFlow
React
Node.js

Tools & Databases

Docker
Git
MongoDB
SQL

Experience

July 2025 - Current

Junior Analyst

Institutional Shareholder Services STOXX

  • Developed a proof-of-concept RAG application on Vespa.ai for document ingestion and retrieval
  • Implemented Python scripts to benchmark & compare Vespa.ai and Elasticsearch performance
  • Enhanced translation system with LLM-based solution and glossary feature
December 2024 - February 2025

Machine Learning Engineer Intern

Mercari Japan

  • Worked in the Trust and Safety team on Item Moderation project
  • Improved LightGBM model precision from 3% to 5%
  • Fine-tuned Japanese BERT models, achieving 6.5% precision
May 2024 - July 2024

Software Engineer Intern

Cere Labs

  • Developed HTML parser for chatbot content ingestion
  • Built service for query-based HTML page extraction
  • Optimized website performance using WebSocket connections

Projects

AI YouTube Search

PyTorch BERT Streamlit Pinecone

A RAG (Retrieval-Augmented Generation) system for YouTube video search using sentence BERT embeddings and vector database.

Credit Card Fraud Detection

Python Flask React Machine Learning

A full-stack application for detecting credit card fraud using machine learning, featuring SMOTE for handling imbalanced data.

Let's connect

Open to collaboration and working.