Divyansh Jain

AI Researcher

Bengaluru, Karnataka, India7 yrs 3 mos experience

AI EnabledAI ML Practitioner

Key Highlights

Expert in machine learning for gaming solutions.
Published research on deep learning applications.
Proven track record in fraud detection and forecasting.

Stackforce AI infers this person is a Data Scientist specializing in Gaming and Fintech industries with strong machine learning expertise.

Contact

Skills

Core Skills

Machine LearningData EngineeringNatural Language ProcessingAi DevelopmentDeep LearningFraud DetectionData AnalysisForecasting

Other Skills

Anomaly DetectionApache AirflowBusiness InsightsC++Customer ServiceData Pipeline AutomationData PipelinesData VisualizationGen-AIGitKubernetesLeadershipMLOpsMatlabMicrosoft Excel

About

As a Staff Data Scientist at MPL, I thrive on leveraging machine learning to drive impactful solutions in the gaming industry. From building predictive models for player actions in Rummy to deploying real-time fraud detection systems, I’m passionate about creating data-driven products that optimize user experience and enhance gameplay. I have a strong foundation in developing and scaling models, particularly using TensorFlow, PySpark, and AWS tools, and have successfully deployed these at scale to handle millions of daily interactions. My previous experience spans across fintech at Bottomline Technologies, where I focused on fraud detection, and global business forecasting at RedHat. Driven by continuous learning, I have published research on deep learning applications in gaming, and I enjoy writing about data science on platforms like Medium. Let's connect to explore how data can transform experiences!

Experience

Mobile premier league (mpl)

Staff Data Scientist

Apr 2025 – Present · 11 mos · New Delhi, Delhi, India · On-site

Skill-Based Matchmaking System
Developed Player Rating models to accurately predict Player skill across different
skill-based games such as Rummy, Gin-Rummy, Ludo, and other casual games, to ensure fair match-making across different entry fees and geographies.
Developed predictive models using classical machine learning and embedding-based neural networks on historical and behavioral game-play to estimate player win probability.
Built scalable data pipelines for real-time matchmaking, serving millions of requests/day with low-latency predictions using Kubernetes and Redis feature store.

Machine LearningPredictive ModelingData PipelinesKubernetesRedisData Engineering

Gameskraft

2 roles

Lead Data Scientist

Apr 2024 – Apr 2025 · 1 yr · Bengaluru, Karnataka, India

Customer Support Navigation using Gen-AI
Automate app help center navigation based on voice query inputs by the players.
Developed and deployed a Gen-AI-powered customer support navigation system leveraging LLaMA and Gemini for natural language understanding and retrieval-augmented generation (RAG) to provide accurate, context-specific responses and enhance user experience in 5 languages.

Gen-AINatural Language ProcessingVoice Query AutomationAI Development

Data Scientist

May 2022 – Apr 2024 · 1 yr 11 mos · Bengaluru, Karnataka, India

𝐑𝐮𝐦𝐦𝐲 𝐆𝐚𝐦𝐞𝐩𝐥𝐚𝐲 𝐒𝐤𝐢𝐥𝐥 𝐌𝐨𝐝𝐞𝐥𝐬
Recommend player's next best action for Rummy, a popular skill based-card game by identifying
and modeling top player behaviour.
Developed Deep Learning models and proposed a novel frame-work SPADENet.
Research paper accepted in IEEE CAI-2024, Singapore titled "𝘚𝘗𝘈𝘋𝘌𝘕𝘦𝘵: 𝘚𝘬𝘪𝘭𝘭-𝘣𝘢𝘴𝘦𝘥 𝘗𝘭𝘢𝘺𝘦𝘳 𝘈𝘤𝘵𝘪𝘰𝘯 𝘋𝘦𝘤𝘪𝘴𝘪𝘰𝘯 𝘢𝘯𝘥 𝘌𝘷𝘢𝘭𝘶𝘢𝘵𝘪𝘰𝘯 𝘧𝘰𝘳 𝘊𝘢𝘳𝘥 𝘎𝘢𝘮𝘦𝘴 𝘶𝘴𝘪𝘯𝘨 𝘋𝘦𝘦𝘱 𝘕𝘦𝘶𝘳𝘢𝘭 𝘕𝘦𝘵𝘸𝘰𝘳𝘬𝘴 (𝘖𝘯𝘭𝘪𝘯𝘦 𝘙𝘶𝘮𝘮𝘺 𝘢𝘴 𝘊𝘢𝘴𝘦 𝘚𝘵𝘶𝘥𝘺)"
𝐌𝐨𝐝𝐞𝐥 𝐃𝐞𝐩𝐥𝐨𝐲𝐦𝐞𝐧𝐭/𝐌𝐋-𝐎𝐩𝐬
Deployed Deep Learning models for Real-time Inference currently serving >2 Million requests/day with <150ms p99 latency.
Batch prediction services deployed using Map-Reduce for data analysis in Data-Lake house.

Deep LearningModel DeploymentReal-time InferenceMachine Learning

Bottomline technologies

Data Scientist

Sep 2020 – May 2022 · 1 yr 8 mos · Bengaluru, Karnataka, India

𝐀𝐂𝐇 (𝐖𝐢𝐫𝐞-𝐓𝐫𝐚𝐧𝐬𝐟𝐞𝐫) 𝐅𝐫𝐚𝐮𝐝 𝐃𝐞𝐭𝐞𝐜𝐭𝐢𝐨𝐧
Anomaly Detection using Self-Supervised Deep Learning on highly imbalanced dataset (Target Label = 0.0001%)
Supervised Fraud Detection to reduce marked False Positives transactions by Bank rule engine. Improved precision to 0.4 with recall = 1
𝐓𝐨𝐩𝐢𝐜 𝐌𝐨𝐝𝐞𝐥𝐢𝐧𝐠 𝐨𝐧 𝐂𝐮𝐬𝐭𝐨𝐦𝐞𝐫 𝐒𝐮𝐩𝐩𝐨𝐫𝐭 𝐓𝐢𝐜𝐤𝐞𝐭𝐬
Unsupervised Topic Modeling on customer support tickets using LDA & BERTopic to reduce resolution time for Customer Support Team.

Anomaly DetectionFraud DetectionTopic ModelingData Analysis

Red hat

Data Scientist

Dec 2018 – Sep 2020 · 1 yr 9 mos · Pune, Maharashtra, India

𝐋𝐨𝐧𝐠 𝐓𝐞𝐫𝐦 𝐑𝐞𝐯𝐞𝐧𝐮𝐞 𝐅𝐨𝐫𝐞𝐜𝐚𝐬𝐭𝐢𝐧𝐠
Developed Hierarchal Time Series Model using ensemble methods (Prophet, ARIMA, Neural Networks, ETS)
Achieved at Par Performance on MAPE as compared to business forecast with <3% error deviation with 1-year future forecast window.
Designed Data Pipeline to automate data aggregation and generate historical snapshots for feature engineering using Airflow.
𝐇𝐞𝐚𝐝𝐜𝐨𝐮𝐧𝐭 𝐅𝐨𝐫𝐞𝐜𝐚𝐬𝐭
Developed Headcount Forecast Model to accurately predict time to close open roles for Talent Acquisition team surpassing business judgement. Improved Median Absolute Error by >3 weeks.

Time Series AnalysisData Pipeline AutomationForecastingData Engineering