Dec 2024 – May 2025 · 5 mos · Remote

• Contributed to backend systems for real-time market data ingestion, aggregation, and distribution across multiple cryptocurrency exchanges using efficient messaging and multiplexing techniques.
• Designed and implemented a unified market data aggregation layer to consolidate and serve live order book snapshots with configurable parameters.
• Optimized key components in low-latency data pipelines, achieving significant reductions in processing time at the microsecond scale.
• Set up a dynamic compute management system on GPU clusters, enabling intelligent switching between compute-intensive tasks based on external triggers.
• Wrote unit tests for core modules including execution logic and external data connectors, ensuring system robustness and maintainability.
• Conducted performance benchmarking of internal data structures, analyzing trade-offs in latency and throughput across various implementations.

May 2024 – Jul 2024 · 2 mos · Gurugram, Haryana, India · On-site

• Built internal command-line utilities for analyzing and monitoring file system patterns, including summarization and anomaly detection.
• Optimized data processing workflows, handling large-scale logs with a focus on performance and responsiveness.
• Developed tools around task orchestration systems to improve reliability, efficiency, and resource management.
• Designed test environments to streamline deployment setups and configuration testing.

May 2023 – Apr 2024 · 11 mos · Remote

• Developed a MultiModal Consistency Preserving Self-Reflective Retrieval Augmented Generation Pipeline for fact verification and validation using advanced models.
• Contributed to developing a fact verification and correction pipeline through evidence generation using Large Language Models. Created various prompting templates for question generation, answering, and keyphrase extraction. Fine-tuned several transformers for classification and correction tasks using advanced techniques.
• Enhanced the interpretability of black-box machine learning models trained with AutoML by developing metrics to evaluate and improve model transparency, accountability, correctness, simplicity, and trustworthiness.

Jun 2020 – Aug 2020 · 2 mos

• Contributed to designing and implementing a web ecosystem for the real estate industry. The web ecosystem consisted of various online tools and platforms that enabled multiple stakeholders, such as agents, buyers, and sellers, to collaborate on documents related to property transactions. The web ecosystem aimed to streamline the buying, selling, renting, and managing of real estate properties.
• Developed the web tools using Node.js, Express.js, Sendgrid, and Firebase functions, firestore, authentication for the backend that included notifications, emails, and regular updates on changes to the document, and the front end was developed using React, and Redux.