Headline

Kaggle Competition Expert(Top 1%) | Senior AI Engineering Lead - Watsonx COE | GenAI | Developed Vibe Coding Platform | LLM | RAG | AI Agents | AI/ML Architect | ML & NLP Expert | DL | MLE | MLOps | Azure | AWS | Mentor

About

Visionary Senior AI Engineering Lead and Technical Manager with 13+ years of experience spearheading end-to-end product design across Generative AI, Agentic AI, and Advanced Machine Learning. Distinguished as a Top 1% Kaggle Competition Expert, with a proven record of transforming complex research into scalable, enterprise-grade AI deployments that drive commercial value and operational excellence. Currently working in the Watsonx Client Engineering COE Team at IBM, where I: - Architected and led development of an enterprise-grade Vibe Coding platform for IBM partners, enabling autonomous generation and deployment of full-stack agentic applications using IBM technologies. Delivered the platform in 1 month with a lean 3-member engineering team. - Lead the Agentic Center of Excellence (COE) and provide technical expertise across the APAC region. - Delivered 25+ enterprise-grade Generative AI solutions using the IBM Watsonx platform. - Built GenAI platforms for automating RFPs, RFIs, and sales research using Agentic workflows. - Mentored and led a team of 30+ AI engineers, focusing on scalable solution delivery and capability growth. - Served as Judge in flagship AI events including the Worldwide Watsonx Challenge and APAC Agentic Hackathon. - Published research on Hierarchical Rerank Retriever, achieving near-perfect relevance and high MRR in document retrieval. - Recipient of the "Client and Partner Success Award – 2024" and "Culture Catalyst Award - 2025" for outstanding contributions to client delivery and strategic collaboration. Previously at ABI, I: - Won the last 3 Global Hackathon competitions, two as a solo participant. - Founded a new vertical to commercialize in-house analytics products, successfully securing external clients and seed funding. - Developed Analytics As A Service (AaaS) infrastructure to secure our IPs while providing our in house products to external client. Previously at Sopra Steria, I: - Developed a Semantic Search Engine using fastText, NLP, and Python, featuring real-time model training via Redis event queues and Huey task scheduling—designed for scalable, low-latency information retrieval. My experience spans Banking, Retail, Healthcare, and Financial Services. I also spent 1.5 years onsite in Uruguay (South America), deepening my understanding of global delivery and cross-cultural collaboration.

Location

Bengaluru

AI stack

anthropic apis

langchain

openai

replit

huggingface

LlamaIndex

Work experience

Senior AI Engineering Lead(Watsonx COE) - Generative AI & Agentic Solutions at IBM, Apr 2024 - Present · 2 yrs 6 mos, Bengaluru, Part of the APAC Top Team, providing technical expertise across markets and leading the Agentic Center of Excellence (COE) to drive GenAI incubation and reusable asset development. Architected and led the development of “Partner Asset Studio” - an enterprise-grade Agentic Application Studio and Vibe Coding platform for IBM partners. Enabled autonomous generation and deployment of end-to-end agentic applications from natural language prompts using IBM technologies. The platform automatically builds backend services, frontend applications, AI workflows, deployment pipelines, and deploys solutions on IBM Code Engine while providing production-ready source code and live application access. Successfully delivered the platform within 1 month by leading a high-performing 3-member engineering team. Delivered 25+ enterprise-grade Generative AI solutions leveraging the IBM Watsonx platform across diverse industries. Leading and mentoring a team of 30+ AI engineers, supporting architecture design, solution delivery, and skill enablement across regions. Built the RFI GenAI Platform to automate analyst responses (e.g., Gartner) for the Watsonx Product Team. Developed the Agentic Proposal Automation Platform, enabling clients to auto-generate tailored responses for complex RFPs using agentic workflows. Created a GenAI-powered product for report generation in the media industry, significantly reducing manual effort and turnaround time. Judge at major AI hackathons including the Worldwide Watsonx Challenge and APAC Agentic Watsonx Hackathon. Published research on Hierarchical Rerank Retriever, an advanced retrieval method with high MRR and perfect hit rate in chunk selection. Recipient of the "Client and Partner Success Award – 2024" for outstanding contributions to client delivery and strategic collaboration.

Senior Technical Manager Data Science - MMM Product at AB InBev GCC India, Oct 2022 - Mar 2024 · 1 yr 6 mos, Bengaluru, Karnataka, India, Created global MMM product which gives Media and Marketing investment recommendations to Marketing team across 25+ Markets and 100+ Brands in 6 zones. Solution is designed to uncover the complex Non-Linear relationship between investment and impact by learning Non Linear ROI curve using economic priors. It has been designed using Tensorflow low level APIs to make it a complete white box solution where we can query the model to get impact of each factor.

AI/ML Solution Architect (Analytics | MLOps) at AB InBev GCC India, Sep 2021 - Sep 2022 · 1 yr 1 mo, Bangalore Urban, Karnataka, India, Given the opportunity to start a new venture for ABI to commercialize our in-house analytics products in the market. Successfully commercialized our one of the products with external client and moved the team from pre seed to seed funding. Scaled the team to 11 member team to support our commercialize initiative. Developed Analytics As A Service(AaaS) infrastructure to secure our IPs while providing our in house products to external clients. Developing End to End products on Azure using Azure ML, Azure DevOps, ADF, Databriks, Snowflake etc. Implementing best practices followed in MLOPs into the products like CI/CD, MLFlow, Data/Model Monitoring, Data/Model Drift, Event Based Orchestration, Loggers with Automated Alerts, Pytest using YAML etc

Senior Data Scientist at AB InBev GCC India, Jan 2020 - Aug 2021 · 1 yr 8 mos, Bengaluru Area, India, Worked as APAC Zone lead and Commercial tower lead where I lead a team to develop scalable products like Cashflow Forecasting, Upsell Recommendation Engine, Credit Risk etc. Worked with Business team to come up with 1YP plan for different countries in APAC. Setup BU ISEA Analytics Capability to provide analytics products across BU ISEA. Got second rank in Global Gacathon and in ML competition (MLJam) conducted across ABInBev analytics team.

Lead Data Scientist at Sopra Steria, Jun 2019 - Dec 2019 · 7 mos, Noida Area, India, Project : INTELLIGENT SEARCH Developed fastText model for Semantic Search Engine which was trained on data available on different portals of over 200 projects. Data included documents (docx, pdf, xlsx, pptx, txt, rtf) of SharePoint, content of WIKI, JIRA and Outlook Calendar. Provided the feature of Real time Over the Top training of fastText model for document add, update and delete using event Queue (Redis) processing and Task Scheduler (Huey – light version of Celery). Used Context based Spell Checker (complete logic developed by me) to correct the spelling mistakes and used tfIdf vectorizer as a booster to get an accuracy of around 90%.

Data Scientist at Tata Consultancy Services, Mar 2013 - May 2019 · 6 yrs 3 mos, Noida Area, India, Project : Custom Entity Extraction from Search Query Implemented NLP search bar on the client dashboard so that instead of manual selection of filters on UI, user can type the query in NLP search box and get the filters set on UI. To solve this business requirement we have used RASA NLU for customized entity training and entity extraction (NER-CRF algorithm) from the NLP query. Other tools which we have used are CHATITO for training data generation. We have created end to end product which included creating model, creating pipeline for production deployment using Flask and Gevent WSGIServer. Project : Sentiment Analysis for Product Reviews Build a model for doing the sentiment analysis of the product reviews on the client data. To solve this, I had used Spacy, tfIdfVectorizer, CountVectorizer, Word2Vec, WordEmbeddings for processing the text data and then feed into different models like LinearSVC, Logistic Regression and Naive Bayes. Project : Build Credit Risk Model For my banking client, we build the model to predict the probability that a particular customer would be a defaulter in paying his/her loans as per the last loan records and payments records. Training data had 2.5 lacs records with 3 different data files. Used different algorithms like KNN, LightGBM, Random Forest and did Stacking of all to solve this requirement. Also, used different pre-processing techniques like FancyImputer and Target Mean Encoding.

Education

Bachelor of Technology - BTech, Electrical, Electronics and Communications Engineering, Gurukula Kangri Vishwavidyalaya

Udacity Nanodegree, Natural Language Processing, Udacity

Public links

github: https://github.com/Ashufet