Jan 2025 – Aug 2025 · 7 mos · San Francisco, California, United States · On-site

• Generated $300K in deal flow from six clients in the mortgage space within 20 days with just a single demo of a multi-agent mortgage pre-approval solution.
• Built over 20 agents across various domains and industries, contributed to the AgentStack framework introducing significant features, for agent development from scratch.

Oct 2024 – Jan 2025 · 3 mos · San Francisco Bay Area · On-site

• Developed four AI agentic solutions for restaurants—“Chat with Restaurant Menus,” an “AR Server Assistant”, a “Waitlist Automation Voice Agent”, and “AI Server Training Agents”—solving operational bottlenecks and improving guest satisfaction.
• Led 40+ customer interviews to refine prototypes, validate the MVP, and achieve product–market fit, securing the first paying client within three months.
• Delivered seamless integrations of NLP, AR, voice, and conversation functionalities that optimized workflows, elevated staff training, and streamlined reservations.

May 2024 – Oct 2024 · 5 mos · New York City Metropolitan Area · Remote

• Developed a patient reidentification system using a large language model with 300M parameters, achieving 90.28% accuracy in identifying patients from redacted clinical notes.
• Collaborated with esteemed Cornell professors to develop a groundbreaking method that successfully reidentified 9% of patients from anonymized medical data, showcasing vulnerabilities in privacy methods.
• Optimized input data size and note selection process, boosting system accuracy from 16% to an impressive 90.28%, significantly improving model effectiveness.

Feb 2024 – Apr 2024 · 2 mos · New Jersey, United States · Hybrid

• Implemented an AutoML pipeline for seamless car parts detection, segmentation, and annotation from noisy images.
• Fine-tuned and optimized three key transformer models, resulting in a 60% increase in IoU score and precision. This led to an impressive 87% reduction in false positives, enhancing the overall accuracy of car parts detection.
• Revamped data preprocessing pipeline by modularizing it, reducing processing time by 40%. This significantly boosted real-time efficiency on local compute systems, improving model performance without the need for external GPU servers.

Aug 2023 – May 2024 · 9 mos · New York, New York, United States · On-site

• Co-authored a research paper (pending approval) in collaboration with Weill Cornell Medicine and Cornell University professors, proposing enhancements to state-of-the-art clinical text de-identification tools using a neural re-identification tool.

Jun 2023 – Aug 2023 · 2 mos · New York, New York, United States · Remote

• Engineered a scalable solution for real-time DeFi analytics dashboards, using Python, GraphQL, Typescript & React to enhance user experience and increase engagement by 40%.
• Implemented machine learning algorithms like K-means and DBSCAN to analyze cross-platform user data, achieving an 85% accuracy in clustering and pattern recognition, leading to more personalized user insights.
• Collaborated with the product team to refine Panoptic protocol's analytics dashboard features based on user feedback, resulting in a 30% increase in daily active users.

Sep 2021 – Jan 2022 · 4 mos · India

• Engineered and deployed a Named-Entity Recognition (NER) model to analyze unstructured medical data, achieving a 70% accuracy rate in diagnosing critical illnesses by fine-tuning a pre-trained BERT-based model.
• Implemented an LSTM-based forecasting algorithm to successfully predict inpatient length of stay and outpatient conversion rates with 85% accuracy over various time-series data sets.

Apr 2021 – Feb 2022 · 10 mos

• FinFlee is an interactive and engaging platform that educates how to invest in the stock market by experiencing the process with the help of a simulated environment.
• Created the backend of an educational trading application using Node.js and Firebase, enabling 2K transactions/month.
• Implemented user-to-user trading algorithms for various trading strategies such as Mean Reversion, Momentum Trading, Arbitrage, High Frequency, and Pair Trading.

Apr 2021 – Jan 2022 · 9 mos · Boston, Massachusetts, United States

• Implemented an XCeption architecture-based video interpretation algorithm to analyze online meeting attendees' emotions and moods in real-time, resulting in 3x faster recognition by reducing frames-per-second captured.
• Used Python and TensorFlow to optimize the video interpretation algorithm, reducing processing time by 50%.
• Developed and implemented OLS regression models with 90% accuracy to uncover trends revealing positive correlations among psychometric qualities within groups, leading to a 25% improvement in team synergy.

Aug 2020 – Aug 2021 · 1 yr · India

• Collaborated with Google Developer (DevRel) Advocates to promote technical awareness among student communities.
• Organized over 50 technical events and hackathons, engaging a digital footprint of over 100k students globally.

Aug 2020 – Aug 2021 · 1 yr

• Formed India’s largest student-run community platform which is now a collaboration of 40+ DSCs all across India.
• The objective of this platform is pretty straightforward- holding regular competitive programming contests and nurturing a strong culture of problem-solving and innovative thinking. We aim to provide enthusiastic coders an opportunity to use their trusty grey cells, solve interesting challenges, and compete against other universities and technical institutions across the country.