Ruchin Dhama, Developer in Bengaluru, India
Ruchin is available for hire
Hire Ruchin

Ruchin Dhama

Bio

Ruchin is a senior machine learning engineer with over five years of experience in recommendation systems and generative AI for companies including Walmart and Samsung. He achieved up to 40% efficiency gains and a 4 – 5% total return on advertising spend at Walmart by deploying production-scale machine learning solutions. His expertise spans TensorFlow, PyTorch, and AWS for the retail and advertising industries.

Portfolio

Walmart Global Tech
LightGBM, ARIMA, Apache Airflow, Generative Artificial Intelligence (GenAI)...
Samsung R&D Institute India (SRIB)
SQL, PySpark, AWS IoT, Big Data, Flink, Apache Kafka, Python 3, Apache Airflow...
Travash Software
U-Net, Machine Learning, Deep Learning, Signal Analysis...

Experience

  • Recommendation Systems - 5 years
  • Machine Learning - 5 years
  • Python 3 - 5 years
  • Generative Artificial Intelligence (GenAI) - 5 years
  • SQL - 5 years
  • PySpark - 5 years
  • Retrieval-augmented Generation (RAG) - 5 years
  • Agentic AI - 3 years

Preferred Environment

AWS IoT, Azure, Google Cloud Platform (GCP), Docker, Terraform, Jenkins, Apache Airflow, Recommendation Systems, Retrieval-augmented Generation (RAG), Python

The most amazing...

...recommendation system I've built delivered a 4 – 5% total return on advertising spend uplift at Walmart.

Work Experience

Software Development Engineer 3 | Machine Learning Engineer

2024 - PRESENT
Walmart Global Tech
  • Designed and deployed a production end-to-end machine learning ranking and serving system for Walmart's sponsored ads platform, delivering 4 – 5% advertiser tROAS uplift and directly growing platform ad revenue.
  • Productionized and maintained daily click and impression forecasting pipelines, reducing forecast error by approximately 10% across $5+ million in managed ad spend.
  • Engineered an unsupervised detection system combining Shannon entropy and probabilistic click modeling to flag anomalous sessions in real time, mitigating bot traffic that was corrupting engagement KPIs.
  • Built two production GenAI systems, including a large language model-based audience ID recommender and an agentic catalog chatbot with contextual ad insertion, reducing campaign setup time by approximately 40%.
Technologies: LightGBM, ARIMA, Apache Airflow, Generative Artificial Intelligence (GenAI), Large Language Models (LLMs), AI Agents, Agentic AI, Forecasting, Spark ML, Recommendation Systems, Retrieval-augmented Generation (RAG), Python, Pandas, Python 3, Git, Time Series Forecasting, Artificial Intelligence (AI), Data Engineering, Finance, LangGraph, Datasets, API Integration, ETL, Quantitative Modeling, Azure Databricks, Data Analysis, Data Analytics, AI Modeling, Amazon Web Services (AWS), GitHub, Machine Learning Infrastructure Engineer, Applied Statistics, Data Visualization, Databases, Statistics, Financial Modeling, Time Series, Data Preprocessing, Large Data Sets, Statistical Analysis, XGBoost, Amazon Bedrock, FastAPI, NumPy, Regex, Scikit-learn, Communication, Data Pipelines, Demand Forecasting, Feature Engineering, Linear Regression, Model Deployment, Model Development, Monitoring, Random Forests, Stakeholder Management, BigQuery, Causal Inference, Google Cloud Storage, Vertex AI, Decision Trees, Documentation, LLM Fine-tuning, LLM Reasoning, Vector Databases, Regression, Applied AI, Hugging Face, PyTorch, Claude, IT Strategy, Kubernetes, LoRa, LangChain, OpenAI SDK

Senior Engineer

2022 - 2024
Samsung R&D Institute India (SRIB)
  • Architected the end-to-end SSP event ingestion pipeline from scratch across millions of events per hour, then optimized it to process 15-minute micro-batch windows, reducing data freshness from 24 hours to 15 minutes.
  • Implemented frame similarity clustering to filter duplicate frames before Amazon Rekognition analysis, reducing API calls by approximately 40 – 50% and saving about $6,000 per year in Rekognition costs while maintaining coverage.
  • Developed an LLM-powered data bot with SQL tool use that automated approximately 80% of recurring manual stats extraction tasks, delivering near-instant responses to queries that previously took hours.
  • Conducted PySpark-based analysis across months of ad-serving event logs, identified failed acknowledgment callbacks from advertiser endpoints as the primary source of metric discrepancies, and shipped an automated reconciliation check to production.
Technologies: SQL, PySpark, AWS IoT, Big Data, Flink, Apache Kafka, Python 3, Apache Airflow, Retrieval-augmented Generation (RAG), CVS, Python, Pandas, Git, Kubeflow, Databricks, Artificial Intelligence (AI), Data Engineering, LangGraph, Datasets, API Integration, ETL, Azure Databricks, Data Analysis, Data Analytics, AI Modeling, Amazon Web Services (AWS), GitHub, Machine Learning Infrastructure Engineer, Applied Statistics, Data Visualization, Databases, Chatbots, Statistics, Time Series, Data Preprocessing, Large Data Sets, Statistical Analysis, Amazon Bedrock, FastAPI, NumPy, Regex, Scikit-learn, Communication, Data Pipelines, Feature Engineering, Linear Regression, Model Deployment, Model Development, Monitoring, Random Forests, Stakeholder Management, BigQuery, Causal Inference, Google Cloud Storage, Vertex AI, Decision Trees, Documentation, LLM Fine-tuning, LLM Reasoning, Reinforcement Learning, Vector Databases, Regression, Applied AI, Knowledge Graphs, Hugging Face, Multimodal Models, Open Weights, Video Analysis, PyTorch, Visual Language Models (VLMs), Claude, IT Strategy, Azure Cognitive Search, Kubernetes, Azure Machine Learning, LangChain

Machine Learning Engineer

2021 - 2022
Travash Software
  • Owned end-to-end machine learning delivery for three products: speaker recognition (audio embeddings), NLP resume parser (greater than 90% entity accuracy), and medical image segmentation (U-Net), enabling Travash to win follow-on contracts.
  • Managed end-to-end project, including client communication and defining application architecture.
  • Collaborated with cross-functional teams, enhancing performance and integrating trained models into the application.
Technologies: U-Net, Machine Learning, Deep Learning, Signal Analysis, Natural Language Processing (NLP), Python, Pandas, Python 3, Git, Artificial Intelligence (AI), Datasets, API Integration, ETL, Data Analysis, Data Analytics, AI Modeling, Amazon Web Services (AWS), GitHub, Machine Learning Infrastructure Engineer, Applied Statistics, Data Visualization, Databases, Chatbots, Statistics, Statistical Analysis, XGBoost, Amazon Bedrock, FastAPI, NumPy, Regex, Scikit-learn, OpenCV, Bayesian Statistics, Communication, Data Pipelines, Feature Engineering, Linear Regression, Model Deployment, Model Development, Monitoring, Random Forests, Stakeholder Management, Causal Inference, LSTM, Decision Trees, Documentation, Reinforcement Learning, Web Scraping, Regression, Beautiful Soup, Applied AI, Hugging Face, Multimodal Models, Open Weights, Video Analysis, PyTorch, Azure Cognitive Search, Azure Machine Learning

Data Science Intern

2021 - 2021
VerSe Innovation
  • Contributed to NLP and computer vision projects by performing data cleaning, feature engineering, and visualization across real-world content datasets.
  • Tracked metrics like views, clicks, watch time, retention, shares, and CTR.
  • Built simple reports and dashboards for the product or content team.
Technologies: Natural Language Processing (NLP), Computer Vision, Python, Pandas, Python 3, Git, Artificial Intelligence (AI), Datasets, API Integration, Data Analysis, Microsoft Power BI, Data Analytics, AI Modeling, Amazon Web Services (AWS), GitHub, Machine Learning Infrastructure Engineer, Applied Statistics, Data Visualization, Statistics, Statistical Analysis, XGBoost, Amazon Bedrock, FastAPI, NumPy, Regex, Scikit-learn, OpenCV, Bayesian Statistics, Communication, Feature Engineering, Linear Regression, Model Deployment, Model Development, Monitoring, Stakeholder Management, Causal Inference, LSTM, Documentation, Web Scraping, Regression, Beautiful Soup, Applied AI, Hugging Face, Open Weights, PyTorch

Experience

Text-to-image Generation

https://github.com/ruchind159/text_to_image
I built a text-to-image generation project using generative adversarial networks (GANs) to generate images from textual descriptions. The project involved training a generator-discriminator architecture to learn the relationship between text embeddings and corresponding visual features, enabling the model to create realistic images conditioned on input text. The solution focused on preprocessing text-image pairs, training stability, and evaluating the quality and relevance of generated outputs.

Face Mask Detection and Social Distancing Monitoring System

https://github.com/ruchind159/MaskDetection
I developed a computer vision-based safety monitoring system that detects whether individuals are wearing face masks and measures the distance between people to identify potential social distancing violations. The project uses Python, OpenCV, Haar Cascades, and deep learning models for real-time face and mask detection through camera feeds. It integrates face detection, mask classification, and distance estimation techniques to provide automated health and safety compliance monitoring. The solution can be applied in public spaces, workplaces, educational institutions, healthcare facilities, and retail environments to improve safety awareness and reduce manual monitoring efforts. The project also demonstrates practical experience in machine learning model deployment, image processing, and real-time video analytics.

Concrete and Pavement Crack Detection Using Deep Learning

https://github.com/ruchind159/crack_detection
I developed an AI-powered crack detection system for identifying structural defects in concrete surfaces, pavements, and infrastructure assets using computer vision and deep learning techniques. The project processes input images to automatically detect and classify cracks, helping improve inspection accuracy and reducing manual effort in infrastructure maintenance. Leveraging Python, OpenCV, and convolutional neural networks (CNNs), the solution performs image preprocessing, feature extraction, and crack localization to identify damaged regions with high precision. This project demonstrates practical experience in defect detection, image segmentation, pattern recognition, and machine learning model development for civil engineering and infrastructure monitoring applications. The system can support preventive maintenance, quality assurance, and structural health monitoring workflows in construction and transportation industries.

Education

2017 - 2021

Bachelor's Degree in Computer Engineering

Dr. D.Y. Patil School of Engineering (DYPSOE) - Pimpri-Chinchwad, India

Certifications

OCTOBER 2020 - PRESENT

Build Basic Generative Adversarial Networks (GANs)

DeepLearning.AI | via Coursera

JULY 2020 - PRESENT

DeepLearning.AI TensorFlow Developer

DeepLearning.AI | via Coursera

JUNE 2020 - PRESENT

Getting Started with AWS Machine Learning

AWS | via Coursera

JUNE 2020 - PRESENT

Deep Learning Specialization

DeepLearning.AI | via Coursera

MAY 2020 - PRESENT

Data Science Math Skills

Duke University | via Coursera

Skills

Libraries/APIs

PySpark, PyTorch, Keras, Pandas, OpenCV, NumPy, Scikit-learn, Beautiful Soup, TensorFlow, Spark ML, XGBoost, LSTM

Tools

Git, GitHub, Visual Language Models (VLMs), Claude, Azure Machine Learning, ARIMA, Terraform, Jenkins, CVS, Flink, Apache Sqoop, Oozie, Microsoft Power BI, BigQuery, Apache Airflow

Languages

SQL, Python, Python 3, Regex, Snowflake, Java, C++

Paradigms

ETL

Platforms

Amazon Web Services (AWS), AWS IoT, Google Cloud Platform (GCP), Docker, Apache Kafka, Databricks, Vertex AI, Kubernetes, Azure, Kubeflow

Storage

MySQL, Databases, Data Pipelines, Apache Hive, PostgreSQL, Database Management Systems (DBMS), Google Cloud Storage

Industry Expertise

Applied Statistics

Frameworks

LightGBM, Hadoop, Flask, Django, LangGraph

Other

Generative Artificial Intelligence (GenAI), Large Language Models (LLMs), Natural Language Processing (NLP), Computer Vision, Recommendation Systems, Retrieval-augmented Generation (RAG), Deep Learning, Machine Learning, Data Science, AI Model Training, Time Series Forecasting, Artificial Intelligence (AI), Data Engineering, Datasets, API Integration, Data Analysis, Data Analytics, AI Modeling, Machine Learning Infrastructure Engineer, Data Visualization, Statistics, Time Series, Data Preprocessing, Large Data Sets, Statistical Analysis, Amazon Bedrock, FastAPI, Communication, Feature Engineering, Linear Regression, Stakeholder Management, Decision Trees, Documentation, Web Scraping, Regression, Applied AI, Hugging Face, Multimodal Models, Open Weights, Geometry, U-Net, Kafka, Paper review, AI Agents, Agentic AI, Forecasting, Big Data, Signal Analysis, Generative Adversarial Networks (GANs), Neural Networks, Time Series Analysis, Fine-tuning, Applied Mathematics, Image Analysis, Transfer Learning, Model Evaluation, Algorithms, Finance, Quantitative Modeling, Azure Databricks, Chatbots, Financial Modeling, Demand Forecasting, Model Deployment, Model Development, Monitoring, Random Forests, Causal Inference, LLM Fine-tuning, LLM Reasoning, Reinforcement Learning, Vector Databases, Video Analysis, IT Strategy, Azure Cognitive Search, LoRa, LangChain, Derivatives, Probability Theory, Calculus, Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Image Processing, Real-Time Video Analytics, Object Detection, Image Segmentation, Structural Health Monitoring, Pattern Recognition, Data Annotation, Trading, Bayesian Statistics, Knowledge Graphs, OpenAI SDK

Collaboration That Works

How to Work with Toptal

Toptal matches you directly with global industry experts from our network in hours—not weeks or months.

1

Share your needs

Discuss your requirements and refine your scope in a call with a Toptal domain expert.
2

Choose your talent

Get a short list of expertly matched talent within 24 hours to review, interview, and choose from.
3

Start your risk-free talent trial

Work with your chosen talent on a trial basis for up to two weeks. Pay only if you decide to hire them.

Top talent is in high demand.

Start hiring