CV

AI Engineer

Apexon

Embedded at Goldman Sachs within the Reconciliation team.

• Shipped LangChain agents that normalize trade records across counterparty formats, grounded in a retrieval index of internal schemas with typed tool-use for field-level validation and canonicalization. Structured-output decoding plus schema-guided retrieval cut field-mapping errors against a prompt-only baseline in ops-team review.
• Developed an unsupervised anomaly detector (streaming Random Cut Forest via rrcf in Python) over high-dimensional counterparty trade feeds due to label sparsity and reconstruction baselines degrade under regime drift.
• Deployed a conversational root-cause analysis system over Snowflake that runs schema-agnostic text-to-query, pulls trade-lifecycle context, and produces grounded summaries of break root causes with resolution suggestions. Tuned retrieval and prompting against analyst-resolved tickets.
• Partnered with multiple clients across industries to design and demo AI-native solutions, including agentic payments orchestration for Mastercard and M11 regulatory protocol digitization for pharma clients.

Quantitative Researcher

WorldQuant LLC

• Deployed 100+ equity alphas through statistical techniques for US Mid Frequency Trading
• Ranked top 10 in the International Quant Championship 2024 (USA)

Student Researcher

Bank of America

• Built end-to-end portfolio optimization using hierarchical reinforcement learning for goal-based investing
• Achieved 18% improvement in annualized returns and Sharpe ratio of 2.3 over Mean-Variance Optimization
• High-level policy uses GRU-based self-attention for stock selection; low-level policy optimizes gains

Software Developer

Cisco

• Scaled Webex Calling APIs with fault-tolerant REST services in Java (Spring Boot) and React
• Integrated ELK Stack and Kibana monitoring, reducing production downtime by 25%
• Modernized B2BUA for cloud readiness with Kubernetes-based auto-scaling and gRPC containers

Software Developer

Insights Software

• Engineered high-performance ODBC and JDBC drivers using C++ (SimbaEngine SDK) and Java
• Spearheaded integrations for Shopify, Stripe, Facebook, and Google Ads data pipelines

PRISM Research Intern

Samsung Research and Development

• Led a team of four developing CNN-driven NLP model for Bixby utterance classification
• Improved computational efficiency by 35% through selective command processing

Masters

NYU Tandon School of Engineering

Applied Mathematics

• Represented NYU in the International Association for Quantitative Finance (IAQF) competition
• Coursework in quantitative finance, stochastic processes, and optimization

Bachelors

BMS College of Engineering

Computer Science

• Published two research papers (IEEE and IJCSE)
• Samsung PRISM research intern during final year

CompNet: A novel Knowledge Graph Embedding Technique for Link Prediction

International Journal of Computer Sciences and Engineering

Proposed a novel knowledge graph embedding technique for link prediction in knowledge graphs.

Real-time Conversion of Sign Language to Text and Speech

IEEE - ICIRCA 2020

Developed a real-time sign language recognition system converting sign gestures to text and speech using computer vision.