Sanjana Sood

First Abu Dhabi Bank Jan'25- Mar'25

• Developed an end-to-end facial recognition framework using AWS services (SageMaker, Lambda, S3, OpenSearch, Jumpstart and SNS) for real-time identity verification.
• Trained and optimized face embedding models (e.g., DeepFace) for high-accuracy recognition, with integration into scalable SageMaker endpoints.
• Implemented image processing and embedding comparison pipeline with automated alerts and logging, ensuring compliance with client’s security and latency requirements.
• Designed a human-in-the-loop audit system using multimodal LLMs (e.g., Qwen-VL, QVQ-72B) to validate recognition accuracy and reduce false positives.

Internal PS Pilot Work Sep'24-Dec'24

• Conducted an in-depth literature review of state-of-the-art RAG architectures, retrieval methods (dense vs sparse), and evaluation approaches to define gaps and opportunities in enterprise use cases.
• Designed a knowledge framework to benchmark RAG systems across key dimensions — including retrieval precision/recall, factual alignment, faithfulness, novelty, and source attribution.
• Developed a taxonomy of evaluation metrics (e.g., ROUGE, BLEU, faithfulness scores, hallucination rate, context overlap) and identified use-case-specific trade-offs between interpretability and performance.
• Outlined future system enhancements including feedback loops, retrieval-grounding alignment, and knowledge graph integration to improve factual robustness and long-term adaptability.

Deutsche Bank Mar'23- Dec'23

• Led and supervised two different data science products within data conditioning workstream.
• Achieved positive client feedback and cross-division adoption of developed products
• Researched emerging industry concepts and integrated knowledge during proof of concept phase
• Developed and enforced adherence to established client procedures
• Collaborated with stakeholders to define project scope and artistic vision

GlaxoSmithKline Sept'22-Feb'23

• Designed and delivered data-driven CX playbook in collaboration with cross-functional teams for new drug launch
• Translated technical jargon into simplified language to effectively convey project details to non-technical personnel
• Suggested actionable improvements to address operational bottlenecks with new and emerging technology
• Created omnichannel measurement framework to enhance business and engagement metrics

• Led several machine learning initiatives involving design, development and deployment of advanced machine learning algorithms to build data-fueled solutions that drive the business closer to its overall goals and objectives (propensity to cancel, propensity to finance, propensity to churn, lead generation etc)
• Instrumental in helping business continue its evolution into an analytical and data-driven culture
• Won UK Digital Experience Awards in 2019 in 'Best use of Data Analytics and insight' category for one of the projects the team delivered.

• Led multiple projects involving analysis of transcriptomics and proteomics datasets with focus on high-dimensional clustering, dimensionality reduction, linear mixed effects models and classification
• Helped with development of pipelines for next generation sequencing data
• Supervised junior member working on inference and classification problems.
• Presented findings at scientific conferences and fielded questions.

• Worked on multiple machine learning projects with focus on Biomarker discovery
• Enabled collaborators to publish in peer-review journals by providing high-quality results and visualisations
• Taught modules and presented at international workshops and seminars (Karolinska Institute (Stockholm), University of Copenhagen, etc).