Sadik Shaikh

About Sadik Shaikh

I’m an AI and Data Science leader with 8 years of experience architecting scalable NLP…

Experience

1. Associate Director - Data Science

   Perfios

   Apr 2024 — Present · Mumbai, IN

   Leading organization-wide initiatives in NLP and Generative AI, with a focus on LLM integration, intelligent document understanding, and platform-level AI capabilities- Led the architecture of a unified SDK integrating proprietary and OpenLLMs for parsing PDFs, images, and text — enabling internal teams to build document parsers rapidly and consistently- Built a low-code GenAI platform to help non-technical users generate accurate LLM-based parsers from just sample documents and ground truth — reducing development time from weeks to hours- Developed an LLM-powered annotation pipeline using a multi-agent system and Judge LLM, combining automation with human feedback to generate high-quality labeled data- Fine-tuned OpenLLMs (Qwen2.5VL, Gemma3) using efficient training strategies, improving accuracy from 50% to 93–95%, with performance comparable to Gemini Flash 2- Optimized inference pipelines through quantization, KV caching, and distillation, resulting in improved throughput and reduced infrastructure costs- Built traditional ML models for Address Quality and Address Scoring, achieving 96% and 94% accuracy respectively- Developed an LLM-based system to classify bank statement narrations and identify transaction types with high precision- Contributed to cross-functional GenAI initiatives, guiding the adoption of LLM-based solutions in QA and internal AI platforms.

Education

• University of Mumbai

  Bachelor of Engineering (B.E.), Computer Engineering

  2013 — 2017

Skills

• C
• Html
• Data Structures
• Python
• Java
• Machine Learning
• Sql
• Numpy
• Mysql
• Object-Oriented Programming (Oop)
• Scikit-Learn
• Postgresql
• Django