Amay Dubey
ML engineer | NLP & Machine learning expert | Generative AI
Bengaluru
7+ years of work experience
Feb 20251 yr 7 mosPresent
Bengaluru, Karnataka, India
Jun 20246 mosNov 2024
Bengaluru, Karnataka, India
◦ Unstructured Data → Structured Data: Spearheaded the development of a fully automated machine learning pipeline, leveraging NLP, Open AI models, and fuzzy matching techniques to standardize and normalize disparate input columns for insurance data. This system improved data consistency across multiple datasets by 90%.
◦ File Classifier: Engineered a highly accurate multiclass classifier using Random Forest to automatically categorize incoming files on the platform using header extraction and processing from multiple file formats
◦ NER: Implemented an advanced Named Entity Recognition (NER) model using spaCy’s en core web trf model, achieving 95% accuracy in extracting organization names and critical dates from email subjects.
Aug 20221 yr 11 mosJun 2024
Indore, Madhya Pradesh, India
◦ ChatBot: Developed a chatbot for solar plants and BESS systems with advanced prompt engineering, intelligent agents, and chain of thought for seamless KPI analysis, enhancing operational decision-making
◦ Fine tuning LLM: Fine-tuning various open-source language models (LLMs) such as LlaMa2 and Falcon to customize and train an in-house model on proprietary data for text-to-SQL generation
◦ BESS: Analyzed data for Battery Energy Storage Systems(BESS) to detect cycles, identify weak modules system for Zinc bromide, Li-ion batteries.
◦ Data Ingestion: Build an end-to-end ML orchestration pipeline for efficient data ingestion and calculation of KPIs; battery efficiency, charge energy, discharge energy etc.
◦ SoH estimation: Developed and implemented a precise empirical State of Health (SoH) model adept at assessing battery health and projecting remaining useful life for electric vehicles and storage batteries. Attained accurate SoH estimations for electric vehicles, enabling precise forecasts of vehicle remaining range.
◦ Warranty Management: Developed warranty management system in tandem with battery OEM’s to predict expiration of batteries warranty.
Mar 20225 mosJul 2022
Gurugram, Haryana, India
◦ Aspect based Sentiment Analysis: Conducted Aspect-Based Sentiment Analysis for various E-commerce clients, including Samsung, LG etc. Delivered valuable insights into customer sentiments and preferences across different aspects such as quality, installation, after-service, etc
◦ API endpoints: Developed and implemented Flask API endpoints for CRUD operations, enhancing database functionality and optimizing data management processes.
◦ Model Optimisation: Implemented strategy to reduce CPU inference time in our prediction pipeline, effectively implementing ONNX and achieving an 85% reduction in overall inference time
Jun 20192 yrs 10 mosMar 2022
Mumbai, Maharashtra, India
◦ Entity Resolution: Developed an Entity Resolution solution for Metlife Insurance, enhancing data accuracy, data deduplication and reliability.
◦ Proof of Concepts: Led numerous Proof of Concept initiatives for different use cases leveraging LLMs,OpenAI API such as text-to-legal/legal-to-text summarization etc, contributing to innovative solutions.
◦ Research: Collaborated on a research paper comparing transformer-based models, assessing similarity scores on a standardized dataset.
◦ Knowledge Graphs: Utilized Neo4j and Python to convert Informatica Mappings into a Knowledge graph, establishing comprehensive data lineage with query capabilities.
Amay’s AI Stack
Education
Bachelor of Engineering - BE, Information Technology
Acropolis Group of Institutions
20154 yrs2019
Science
St.Norbert Senior Secondary School
20096 yrs2015
Join Amay and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal Boss