Gaurav S M

Lead AI Engineer with 10+ years of experience building AI products, machine learning systems, and customer analytics solutions. Expertise in GenAI, AI Agents, causal inference, recommendation systems, and scalable cloud architectures. Proven track record of delivering production-grade solutions with measurable business impact across marketing, loyalty, and retail domains.

Lead AI Engineer at Anko (2024-03 – Present)

• Developed a GenAI-powered email template generation solution that automates campaign content creation and optimization for marketing teams, including Product matching (via Retrieval & Generation). Technologies: RAG pipelines, cromaDB, AWS Bedrock, Google OR-Tools.
• Architected and delivered an AI-powered marketing platform enabling marketers to create, manage, and deploy campaigns at scale using AWS Bedrock, FastAPI, React, PostgreSQL, Docker, and Kubernetes.
• Designed evaluation criteria and human review workflows to assess LLM-generated content quality, relevance, and business alignment, driving iterative prompt improvements.
• Built an inventory accuracy solution using stacked machine learning models to reconcile robot counts with book inventory. Improved stock accuracy, contributing to an estimated $24M in annual revenue uplift. Technologies: Stacked Ensemble Models, Multi-Layer Perceptron (MLP), CatBoost
• Designed and implemented measurement frameworks to evaluate campaign effectiveness and quantify incremental business impact. Techniques: A/B Testing, Synthetic Controls, Causal Inference
• Led solution architecture and technical design for production-grade AI platforms serving marketing use cases.

Senior Data Scientist at Anko (2022-04 – 2024-03)

• Developed a probabilistic churn prediction model using BG-NBD to jointly estimate customer churn risk and future purchase behavior, generating $2M in incremental revenue and retaining over 25,000 customers.
• Built a CatBoost-based customer classifier to identify early-stage mothers from shopping behavior, enabling targeted marketing campaigns and uncovering a segment of 750K customers.
• Developed a Next Best Category recommendation engine using Word2Vec embeddings to predict customer affinities and drive personalized product recommendations.
• Designed bargain hunter segmentation models using K-Means and Gaussian Mixture Models (GMM) to identify price-sensitive customers and optimize offer personalization.
• Built and maintained end-to-end campaign measurement frameworks to quantify marketing effectiveness and incremental impact using experimentation and causal inference techniques.

Data Scientist at Anko (2021-06 – 2022-03)

• Was responsible for forecasting container volumes to different parts of Australia. Created a unique method of forecasting, which resulted in $26.5M in savings. Used a combination of LightGBM and Arima model
• Developed full-stack application for sourcing team which could help detect product that were faulty during inspection

Senior Associate at Merkle (2020-02 – 2021-06)

• Built Look-alike audiences for multiple companies across different verticals (using 3rd party data, which encompasses information of around 270 MM US population), which helped achieve a large audience from smaller segments to create reach for advertisers.

Decision Scientist at Musigma (2015-08 – 2018-09)

• Built a risk assessment framework (Probability-Severity-Exposure model) to identify regions where ceasing the distribution of marketing paper flyers/pamphlets will have minimal impact, resulting in an estimated savings of $40M

Engineering – Sri Jayachamarajendra College of Engineering, Mysore, India (2011 – 2015)