Aritra Biswas

AI Researcher

Bengaluru, Karnataka, India8 yrs 11 mos experience

Most Likely To SwitchAI ML Practitioner

Key Highlights

Over 8 years of experience in machine learning engineering.
Expertise in building scalable recommendation systems.
Proficient in cloud technologies and ML infrastructure.

Stackforce AI infers this person is a Machine Learning Engineer specializing in MarTech and SaaS solutions.

Contact

Skills

Core Skills

Artificial Intelligence (ai)Applied Machine LearningMachine LearningPython (programming Language)Cloud ComputingAlgorithm DevelopmentData ScienceMarket Research

Other Skills

AlgorithmsAnalyticsApplication DevelopmentBusiness AnalysisC (Programming Language)C++Data AnalysisData AnalyticsData ArchitectureDeep LearningDevOpsDistributed SystemsDockerFlaskGenerative AI

About

As an experienced machine learning engineer with over 8 years in the industry, I specialize in Recommendation Systems, LLMs, Marketing Analytics, ML infrastructure, and MLOps. My expertise lies in Distributed computing and High-Performance Computing (HPC), with extensive knowledge in Distributed model training, Reinforcement learning, and Embarrassingly parallel job execution. I possess proficiency in auto-scaling, infrastructure selection, and optimization, and have a solid background in designing and implementing large-scale machine learning systems. My skills include building efficient and scalable architectures for ML workflows and life cycle. I am dedicated to optimizing system performance and maximizing resource utilization while ensuring reliability and stability in machine learning workflows. I am proficient in programming languages such as Python, R, C, and C++, with hands-on experience in cloud technologies such as Azure Machine Learning, Azure Data Bricks, GitHub actions, Azure Kubernetes Service, Azure Functions, and Azure DevOps. Currently, I am leading a team of machine learning engineers at AB InBev.

Experience

8 yrs 11 mos

Total Experience

4 yrs 5 mos

Average Tenure

5 yrs 11 mos

Current Experience

Ab inbev

4 roles

Senior Manager Analytics

Promoted

Oct 2023 – Present · 2 yrs 8 mos

Artificial Intelligence (AI)Algorithm DevelopmentApplied Machine LearningSoftware DevelopmentGenerative AI

Senior Machine Learning Engineer

Sep 2022 – Sep 2023 · 1 yr

As an ML engineer lead, my primary responsibility is to build a Python ML library curated for recommendation algorithms that cater to B2B recommendation use-cases. This entails designing, implementing, testing, and maintaining multiple algorithms that can provide accurate and efficient recommendations. I prioritize scalability, data preprocessing, model selection (backtesting/walk forward optimization, HPO, AutoML), and performance optimization while staying up to date with the latest research and industry trends.

Python (Programming Language)Machine LearningResearchProgrammingData Science

Senior Data Scientist

Promoted

Sep 2021 – Sep 2022 · 1 yr

As a Senior Data Scientist, I have developed and implemented a cutting-edge methodology and Python SDK using cloud technologies to run automated Marketing Mix Modeling (MMM) at scale. Through this process, I have created a multi-objective marketing budget allocator that enables businesses to identify the optimal media spending strategy.
To ensure seamless integration with existing products, I streamlined and automated the data pipeline using Azure. Through these efforts, I have helped businesses achieve data-driven decision-making and improved ROI on their marketing investments. My expertise in data science has allowed me to contribute significantly to the growth of organizations by leveraging the latest technologies and methodologies.

Machine LearningKubernetesProduct DevelopmentOptimizationCloud Computing

Data Scientist

May 2020 – Aug 2021 · 1 yr 3 mos

My responsibilities included deploying a Linear Mixed Modeling (LMM)-based MMM tool to the cloud and fine-tuning it for operational efficiency. I also collaborated with the MIT BudLabs team to explore and enhance existing MMM methodologies. Furthermore, I have worked on a cloud-based Lift solution that measures in-store impact due to digital campaigns, and I have increased its efficiency by integrating an LLVM-compiled Dynamic Time Warping (DTW) distance algorithm with Sakoe-Chiba band using Numba.

AnalyticsMarket ResearchOptimizationCloud ComputingAlgorithm Development

Nielsen

2 roles

Data Scientist

Promoted

Jul 2018 – May 2020 · 1 yr 10 mos

In this role as data scientist, I was responsible for development of Nielsen StoryBoarder, one of the largest shared services in the marketing effectiveness organization, serving $95MM worth of business across Lift & MMM. I also developed a response curve visualization and aggregation framework to understand the confounding effect of media execution. In the same role I have worked on Rapid MPA (Lift Solution) which was used for one-to-many store matching, synthetic control group generation, and measuring in-store lift due to promotional activities. Scaling DTW for daily store level data was a huge computational challenge in this project, which I solved during project multiplier in ABI. I implemented estimating ad-effectiveness using geo-experiments in a time-based regression framework. I also minimized the runtime of SCM to less than a minute by designing an optimized LLVM compiled code using Numba and Intel MKL.

Python (Programming Language)Market ResearchAlgorithmsData ScienceMicrosoft Power BI

Associate Data Scientist

May 2017 – Jul 2018 · 1 yr 2 mos

As an Associate Data Scientist, my main responsibility was to develop algorithm for to the development of Rapid Modeler and Rapid Simulator - two tools that utilized hierarchical additive linear model methodology and customized Monte Carlo grid search, respectively, for portfolio level marketing budget allocation with vehicle level response curves.
My work on the Rapid MMM Suite included the development of the MMM Modeler and Simulator, with plans to integrate the suite with other Nielsen products. I also developed batch-based Monte Carlo grid search for constraint-based budget allocation for multi-product scenarios and implemented/enhanced SLSQP for budget allocation across marketing tactics for single product scenarios.
To ensure production-ready performance, I developed Numpy and Numba code for simulation using Intel Math Kernel and Short Vector Math Library. Additionally, I developed the ETL process for the modeling process and automated data extraction for marketing planner.
I also collaborated with UI/UX teams, vendor partners, clients, users, and other stakeholders for requirements capturing. Overall, my work contributed to sales and spend optimization, and I deployed the ML code as a REST API endpoint on RHEL servers using Flask and Python while conducting stress testing using Locust.

REST APIMachine LearningMicrosoft AzureDockerAlgorithm Development