Biswarup – Python, Machine learning, Pandas

A managed audience recommendation service for advertisers that lets them expand a seed audience into a larger lookalike audience with similar characteristics. It was built on a heterogeneous identity graph that resolved and unified more than 200M consumer profiles across multiple data providers, replacing a legacy third-party identity solution. The main features were cross-provider identity resolution, graph-based lookalike recommendations powered by a GraphSAGE model, and household-level audience expansion that kept audience quality intact as reach grew.

• Led ML R&D for the service end-to-end
• Architected the heterogeneous identity graph that unified 200M+ consumer profiles across multiple data providers
• Designed and trained the GraphSAGE-based model that generated the audience recommendations
• The platform replaced the primary client's legacy third-party identity solution and expanded their addressable reach from 21M to 85M households (roughly 4x) while maintaining audience quality.

A multimodal AI system for sales teams that analyzed and scored recorded sales calls. It processed thousands of calls a day and ran on proprietary LLMs that were continuously improved with newly labeled data. The main features were automated call scoring, multimodal RAG on call content, and an internal data-labeling platform that fed the model-improvement loop.

• Created the ML infrastructure from the ground up to enable call scoring capabilities at scale; the infrastructure, hosted on AWS, scaled to thousands of calls per day.
• Spearheaded the development of RAG-based agents that powered ReplayzIQ.
• Built a retrieval system that was 2x more capable of finding quality call snippets than standard retrieval using OpenAI embeddings.
• Created internal tools for validation of results and automated reporting of unusual results.
• Created the continuous deployment pipeline.
• Brought seamless human-in-the-loop processes through customization of AWS SageMaker labeling jobs and gathered data for fine-tuning an in-house reranker model and LLM model.
• Fine-tuned an LLM for scoring calls based on annotated data using Mixtral 8-7b on Modal.

A leading global professional services company that provides a broad range of services in strategy, consulting, digital, technology, and operations.

• Built a recommender system for a coffee brand, generating over 1 million in incremental revenue over random assignment
• Developed an experimentation platform to evaluate ML models across customer cohorts, accelerating model iteration cycles.
• Optimized real-time inference for drive-through recommendations, reducing latency and deployment cost using ONNX RT in lambda
• Evangelized data-driven practices across teams via workshops and training sessions

A major acute-care teaching hospital located in Montreal, Quebec, Canada. The hospital is a cornerstone of the integrated health network and is known for its clinical research, medical innovation, and open-door community policy.

• Implemented a next-gen clinical data repository, using a FHIR server to enhance data interoperability.
• Built a COVID-19 severity prediction model to aid clinicians in triage.

A leading global provider of clinical research services, commercial insights, advanced healthcare analytics, and technology solutions for the life sciences and healthcare industries.

• Transitioned traditional analytics to a PySpark and Airflow-based automated analytics pipeline, cutting analysis time from 6+ hours to under 30 minutes.
• Developed an LSTM-based model to detect rare diseases from EHR data, which was integrated via a Flask-based API.