Otkrist Gupta

• Led AI strategy and execution across customer engagement, risk, and operations at a multi-billion-dollar financial services company • Building voice-based conversational AI agents for customer engagement and collections • Established and led specialized ML teams (PhDs + senior engineers) delivering production-grade models and automation for core business processes • Owned development of financial risk scoring models used in underwriting and servicing $2B+ in auto loans and supporting $1B+ in securitized risk portfolios • Architected the in-house MLOps platform, enabling automated model training, optimization, deployment, monitoring, testing, governance, and regulatory auditing • Designed and led development of an internal AutoML platform for scalable hyperparameter optimization and model experimentation across teams • Leading computer vision initiatives for ID verification, bank statement parsing, MRZ extraction, image forensics, and fraud detection • Published peer-reviewed research in applied computer vision and signal processing (IJDAR 2021, EURASIP 2022)

Pursuing exciting projects and goals for LinkedIn recruiter backend.

Working on capture imaging and visualization of light transmission using femtosecond light source and pico second imager. Gave the mathematical proof of existence and uniqueness of solution for around the corner 3D shape problem. Invented algorithm for around the corner 3d imaging using time of arrival data of photons (patent submitted). Publication in Nature Communication (2012), Optics Express (2013) and thesis topic.

Developed trading, market data acquisition infrastructure for trading on LSE, NYSE Euronext (Netherlands), EUREX (Switzerland), BATS, LME (London Metal), SIBE (Spain), and other notable exchanges. All programming was done in C++ Designed and developed using python, a web based risk management GUI to control risk limits for trading Developed a library of scripts to track positions, check database integrity, send alerts, and debug server messages

Automated re-annotation of web pages after html structure modifications using dynamic programming based techniques Developed a robust mechanism for annotation transfer across two different DOMs between web pages Virtual Reality Applications Research Team (VIRART), University of Nottingham (May 2007-July 2007) Member of "Virtual Klub" project team which aimed at developing software aids to impart music education to children Developed a software package in Java which provides interactive 3D environment with attachable music themes Performed regular interaction with school children and user studies to make software aids more user friendly Academic Projects Recovering 3D shape Around a Corner using Ultra -Fast Time-of-Flight Imaging

Worked on software to impart music education to kids. Project included 3d visualization of music, direct interaction and user studies for improvement.

Thesis Title: "Unlocking the potential of deep neural networks in resource and data constrained environments"

Selected for Research Assistantship (2010-2012) with Professor Ramesh Raskar in Camera Culture group Coursework - Advanced Algorithms, Machine Learning, Optimization Methods, Computational Photography Main research area: Novel imaging techniques using time of flight methods, futuristic 3D displays and projectors