Deal Engineering @ Altius, a boutique PE-advisory firm working with funds and their portfolio companies in the Mid-Market.

• Building a contract extraction pipeline to extract contract monitoring rules, figures, and relevant metrics from large-scale, complex contract documents.

• Implementing functionalities for field suggestions, with human review, audit logging, and re-extraction jobs to keep extracted contract data accurate and compliant over time.

Serving as a Course Assistant for the following courses at the university:

• Natural Language Processing, under Prof. Adam Meyers (Spring '25, Fall '25, Spring '26)

• Foundations of Causal Inference in Data Science, under Prof. Darryl Reeves (Fall '25)

Leading weekly recitations, evaluating course assignments, setting and grading exam papers, clarifying doubts, and providing guidance and feedback for student projects.

• Built an internal data analytics platform for the company using Go, and added new functionalities to integrate data from multiple sources, including Snowflake and PostgreSQL.

• Architected and implemented a system to convert natural language questions into executable SQL queries with automated visualizations and graphs using Python and LLMs, enabling non-technical users to build dashboards independently.

• Unlocked $36,000 in yearly cost savings through the development and enhancement of this in-house analytics platform.

• Worked at IISc's Center of Data for Public Good, on implementing differential privacy in real-world use cases in the agricultural and medical domains.

• Designed a synthetic data generator for large low-dimensional datasets, which captured the distribution of the original surveyed data while ensuring that the generated data was differentially private.

• Developed a software solution to compute differentially private statistics and histograms from a dataset. The solution also implemented different DP-adjacent anonymization techniques to release an anonymized dataset. The overall solution was deployed to over 50 metropolitan bodies across the nation.

• Worked in the Global Fraud Solutions team on building a model to flag suspicious user phones, preventing account takeovers during login events.

• Developed a rule-based approach that improved the existing process, leading to 21% more phones being found. Designed and optimized a pipeline for the end-to-end deployment of the model, reducing running time by 88%, and also ensuring data consistency and model explainability.

• Prevented losses of an average $2 million/month by effectively identifying suspicious activity.