Himanshu Srivastava — AI Researcher

As a GenAI, Agentic, , Machine Learning and Deep Learning expert with over 13+ years of experience, I have worked on several challenging projects ranging from Image Classification, Object Detection, Visual Search, Automatic Look Creation, Text Summarization, and more. I have a strong command over Python and OpenCV programming, and I am also skilled in C/C++ and Data Structure on the Linux platform. My expertise lies in developing apparel classification and detection models using Convolution Neural Networks and Yolo Framework. I was responsible for the end-to-end development of these projects, from scratch to production. I have also worked on developing a Visual Search system using Spark and MemSql, which involved optimizing the system architecture and implementing data automation using Spark standalone cluster mode. As a result of my optimization efforts, the final result time was reduced from 11 seconds to just 1 second. I have also worked on developing an Automatic Look Creator using Data Mining concepts of Fp-Growth and a Recommendation System that suggests the most popular products based on events captured in Google Analytics data. Additionally, I have worked on Video Analytics projects, including developing an attendance system by identifying individuals using detection and making entries in a Neo4j database using multiple cameras. In terms of Natural Language Processing, I have worked on several projects such as Text Summarization, Text Classification using Deep Learning Networks, Name and Entity Recognition, and Message Classification models like promo, bank, personal, etc. I have also worked on Computer Vision projects such as Physical Distancing Model, Pothole Detection Model, and Criminal Profiling and Crime Prediction models. Beyond technical expertise, I combine leadership in strategic planning, decision-making, end-to-end delivery ownership, and cross-functional collaboration. I excel at aligning business goals with technology, engaging stakeholders, and driving teams to deliver measurable impact. With a proven record of optimizing architectures, accelerating system performance, and developing data-driven products from scratch to production, I aim to lead innovative AI solutions that transform business challenges into real-world value.

Stackforce AI infers this person is a Data Science and AI expert with extensive experience in Retail and E-commerce sectors.

Location: Gurugram, Haryana, India

Experience: 15 yrs

Skills

Data Science
Genai
Machine Learning
Computer Vision
Nlp
Recommender Systems
Image Processing

Career Highlights

Expert in Machine Learning and Deep Learning.
Proven record in optimizing system performance.
Led innovative AI solutions transforming business challenges.

Work Experience

Accenture

Principal Data Scientist Manager (3 yrs 10 mos)

Data Science Manager (8 mos)

Larsen & Toubro

Assistant Manager - Data Scientist (1 yr 10 mos)

Gupshup

Lead Artificial Intelligance Engineer (6 mos)

STYFI - Your Shopping Companion

Lead Architect - Artificial Intelligence (1 yr 6 mos)

Computer Vision Lead (1 yr)

Home Craft Online Pvt Ltd

Senior Software Engineer (1 yr 4 mos)

The Tata Power Company Limited - Strategic Engineering Division (Tata Power SED)

Senior Software Developer (3 mos)

Paletly Designs Inc

Image Processing Engineer (8 mos)

Indian Institute of Technology, Guwahati

M.Tech in Signal Processing (2 yrs)

Robosapiens India

Research Engineer (1 yr 5 mos)

Education

Master of Technology (M.Tech.) at Indian Institute of Technology, Guwahati

Bachelor's Degree at JSS Academy of Technical Education

Himanshu Srivastava

AI Researcher

Gurugram, Haryana, India15 yrs experience

Most Likely To SwitchAI Enabled

Key Highlights

Expert in Machine Learning and Deep Learning.
Proven record in optimizing system performance.
Led innovative AI solutions transforming business challenges.

Stackforce AI infers this person is a Data Science and AI expert with extensive experience in Retail and E-commerce sectors.

Contact

Skills

Core Skills

Data ScienceGenaiMachine LearningComputer VisionNlpRecommender SystemsImage Processing

Other Skills

Large Language Models (LLM)Prompt DesignMicrosoft AzureTeam ManagementPrompt EngineeringCampaign StrategiesTeam LeadershipGenerative AIMLOpsDockerVector DBPythonProject ManagementChatGPTAzure services

About

Experience

15 yrs

Total Experience

1 yr 8 mos

Average Tenure

4 yrs 6 mos

Current Experience

Accenture

2 roles

Principal Data Scientist Manager

Promoted

Jul 2022 – Present · 3 yrs 10 mos · Gurugram, Haryana, India · Hybrid

Middle East Retailer: Agentic Solution
Multi-Agent System Architecture: Designed and implemented production-grade multi-agent systems with orchestrator, super, and utility agents.
Intelligent Agent Coordination: Enabled agents to intelligently trigger sub-agents based on user inputs and business logic.
ML Integration:Integrated ML models (e.g., Markdown, Assortment) for automated, data-driven decision-making.
Cloud-Native Deployment: Built modular, scalable architecture using cloud-native tools and microservices.
System Reliability & Optimization: Ensured reliability through robust monitoring, fault-tolerant design, and performance tuning.
Large Global Retailer: Conversational Commerce Chatbot (GenAI)
Implemented Azure services, including Blob Storage, Pipelines Repository, and Azure App Service, ensuring smooth development and deployment processes.
Leveraged RAG architecture to enhance the accuracy and relevance of generated content.
Managed the project lifecycle from initial concept and architecture to execution, ensuring successful delivery.
Oversaw technical and functional components, including developing and disseminating business logic to the team.
Utilized Python for programming, PostgreSQL for vector databases, and OpenAI for GenAI capabilities, demonstrating proficiency in diverse technologies.
Implemented necessary modifications following established procedures, ensuring flexibility and adherence to evolving requirements.
Global Oil & Gas Retailer: Customer Analytics
Managed end-to-end delivery including reporting, model development, and in-depth analysis.
Designed campaign management strategies tailored for loyalty customers.
Developed a churn prediction model and implemented customer segmentation techniques.
Constructed customer network graphs and conducted dynamic tiering analysis to gain insights into customer behavior.
Utilized AWS cloud services for both model development and deployment.

Large Language Models (LLM)Data SciencePrompt DesignMicrosoft AzureGenAITeam Management+10

Data Science Manager

Nov 2021 – Jul 2022 · 8 mos · Gurugram, Haryana, India · Hybrid

Indian Retailer: Customer Analytics
Developed an apparel classification model using CNN deep learning and an algorithm for auto-tagging based on images to detect product attributes like category, color, print, and pattern.
Created a machine learning-based outfit recommendation system to enhance customer experience, utilizing attributes such as color, category, customer profile, and interests.
Designed and developed a reverse image search solution using deep learning algorithms to identify and retrieve similar products based on a given image.
Developed a personalized recommendation system based on users' behaviors, activities, and preferences to enhance user experience.
Designed and implemented campaign management strategies for Loyalty customers.
Developed a churn prediction model and customer segmentation techniques.
Constructed customer network graphs and conducted dynamic tiering analysis to gain insights into customer behavior.

Large Language Models (LLM)Data ScienceMachine LearningPrompt DesignDelivery leadGenAI+9

Larsen & toubro

Assistant Manager - Data Scientist

Jan 2020 – Nov 2021 · 1 yr 10 mos · Hyderabad, Telangana, India

Worked on various video analytic, NLP, and data analysis algorithms.
Physical Distancing During COVID19, Build a Physical Distancing Model
which enable agencies to detect the hotspots in the city, where people are coming close to each other against the set social distance. (CNN model and image)
Transformation).
Pothole Detection Build a pothole detection model, which has been a major
cause of road accidents and traffic congestion (combination of ML and image Processing Technique).
Criminal profiling Identify similar criminals based on different features which helps law enforcement agencies
Crime Prediction Build a crime prediction model that helps law enforcement agencies handle incidents and provides first responders with contextual information.

Data ScienceMachine LearningTeam ManagementComputer VisionNLPTeam Leadership+7

Gupshup

Lead Artificial Intelligance Engineer

Jun 2019 – Dec 2019 · 6 mos

Worked on various text classification and NER models.
Classification: build a text message classification model like promo, bank, personal, etc. (LSTM model).
Name and entity recognitionExtract keywords from a given message, like mobile,amount, account number, promo code, etc. ( NER model).

Data ScienceMachine LearningTeam ManagementNLPTeam LeadershipLong Short-term Memory (LSTM)+5

Styfi - your shopping companion

2 roles

Lead Architect - Artificial Intelligence

Dec 2017 – Jun 2019 · 1 yr 6 mos

Online Shop for a Curated Marketplace
Developed a personalized online shopping experience, akin to window shopping.
Recommendation System: Implemented a recommendation system to showcase the most trending or popular products based on events captured in Google Analytics data.
Automatic Look Creation: Developed similar-looking product generations using machine learning solutions (FP-Growth Algorithms).
Visual Search: Conducted research, design, and development for a reverse image search problem, utilizing deep learning feature extraction and Apache Spark for a distributed system.
Text Summarization: Created a text summarization tool using NLP (LSTM - Deep Learning Architecture).

Neo4jData ScienceRecommender SystemsMongoDBTeam ManagementNodejs+6

Computer Vision Lead

Nov 2016 – Nov 2017 · 1 yr

Detection Problem developed an apparel detection model using the Yolo Framework. Did POC and trained the model for apparel detection.
Classification Problem developed an apparel classification model using a convolutional neural network and deep learning. Research and developed an algorithm.
for auto-tagging based upon given images that detect product attributes
like category,colour, print, and pattern, etc.
Video Analytics Developed and is responsible for the attendance system for the office based upon face detection on a real-time camera.

Data ScienceMachine LearningConvolutional Neural Networks (CNN)MongoDBTeam ManagementNodejs+4

Home craft online pvt ltd

Senior Software Engineer

Jul 2015 – Nov 2016 · 1 yr 4 mos · Mumbai Metropolitan Region

The company engages in the online sale of home furnishing products.
Lead flooring project from its development to integration using projections
and transformation of images.
Developed and integrated a tool for wall painting. It includes some image processing operations such as background subtractions, segmentation, and edge
detection and morphological operations.
Responsible for product processing for aggregate products
Developed a tool for a visual search algorithm using machine learning.

Data ScienceMachine LearningConvolutional Neural Networks (CNN)Team ManagementJavaScriptPython (Programming Language)+3

The tata power company limited - strategic engineering division (tata power sed)

Senior Software Developer

Mar 2015 – Jun 2015 · 3 mos · Mumbai Metropolitan Region

Worked in real time Multiple object tracking
Developed a tool for multiple object tracking in real time. The algorithm
involving background subtraction, object classification, and feature extraction.
Implementation of a machine learning algorithm that can detect humans
Manage the complete image processing project before release.

Data ScienceMachine LearningImage ProcessingC++LinuxPython

Paletly designs inc

Image Processing Engineer

Jul 2014 – Mar 2015 · 8 mos · Mumbai Metropolitan Region

Developed image recognition tools to suggest fashion attire and similar products worn by a model in an image.
Developed an algorithm for product classification.
Development of an image processing algorithm involving background subtraction, image segmentation, feature extraction, and pattern recognition.
Implementation of a colour histogram algorithm that can show similar colours present in an image.
Implementation of a machine learning algorithm that can detect different attire.

Machine LearningC++LinuxOpenCV

Indian institute of technology, guwahati

M.Tech in Signal Processing

Jul 2012 – Jul 2014 · 2 yrs · Dispur, Assam, India

Backward-forward reasoning in multiple object tracking : Object tracking is a major issue when it is moving and gets occluded. The thesis is basically an implementation of multiple object tracking with all its reasoning, such as occlusion, i.e., isolation, merging, and splitting into multiple objects.
Isolated word recognition using vector quantization: The objective of this project was to identify the digit from given speech data.