Masters of Science - Computer Science

2023 - 2025

State University of New York at Buffalo, NY - USA

CGPA : 3.7/4.0

Focus: Research Track [AI Focus Area]

Capstone Project: Toxic In-Game Voice Chat Moderation using Multimodal LLMs.

Relevant Subjects: Analysis of Algorithms, Data Intensive Computing, Distributed Systems, Machine Learning, Computer Security, Deep Learning, Operating Systems, Reinforcement Learning, Computer Vision.

Bachelor of Technology - Computer Science

2017 - 2021

Kalinga Institute of Industrial Technology, Bhubaneswar - India

CGPA : 3.95/4.0

Activities: Mozilla Web Development Club.

Lineaje Inc

May 2024 - Present

Software Engineer (Full-time)

• The enterprise platform required a scalable, autonomous interface to handle complex user interactions. Engineered a context-aware multi-agent system orchestrating distinct autonomous agent personas using React, TypeScript, FastAPI, and AWS with session-based memory. Successfully transitioned the platform to an AI-driven experience powered by collaborative agent networks.
• Policy enforcement across massive codebases required autonomous, high-throughput remediation loops. Developed parallelized agentic workflows utilizing n8n and MCP (Model Context Protocol) servers to autonomously detect violations and reason through complex remediation steps. Scaled automated agentic remediations to 1M+ developers while reducing end-to-end processing time by ~18 minutes.
• CI/CD security governance needed to advance from static scanning to active, autonomous fixing. Architected an agentic AI governance scheduler that autonomously evaluates PRs, writes compliant code fixes using LLMs, and pushes remediation PRs automatically. Achieved 90%+ automation of manual security reviews, with autonomous agents enforcing compliant fixes in under 120 seconds.
• Autonomous agents required optimized, low-latency reasoning engines for real-time policy evaluations. A/B tested localized models and migrated to a Dockerized vLLM deployment on AWS EC2, specifically optimized for batched, concurrent agent queries. Reduced agentic reasoning latency from 5 minutes to 10 seconds (~96.7% improvement).

University at Buffalo

Jan 2024 - May 2024

Graduate Research Assistant (Part-time)

• Policy-to-code remediation workflows lacked scalable intelligence. Researched and developed a scalable, context-aware LLM-RAG pipeline. Optimized enterprise compliance and code remediation workflows.
• Enterprise environments required clarity on local vs. cloud AI deployments. Evaluated performance benchmarks for LLM deployments focusing on latency and cost. Delivered strategic compliance and scalability insights for enterprise architecture.

Dell Technologies

2022 - 2024

Software Engineer 2 - IT

• Error logging architecture lacked high-throughput processing resulting in data loss. Architected a scalable PySpark streaming data pipeline extracting >100K error logs daily from PostgreSQL to MongoDB. Enhanced data accuracy and availability by 85%.
• High-priority issues required immediate visibility for stakeholders. Implemented a notification service using Java, ReactJS, RabbitMQ, and Kubernetes. Achieved 60.2% CSAT in FY24 Q2 and reduced errors/week by 1.5x.
• The post-order customer journey suffered from repetitive online UI issues. Innovated a Kappa architecture product using Java, NodeJS, and MongoDB. Reduced customer agent effort by 98%.
• Customer feedback lacked actionable semantic context extraction. Integrated LLMs, RAG vector databases, and RoBERTa. Improved the post-order customer journey to 78.2% CSAT.
• API latency was bottlenecking 100K+ daily records. Developed asynchronous map-reduce services and performed indexing. Increased API speed by 80% and auto-refreshed UX features in React.

Dell Technologies

2020 - 2022

Software Engineer 1 - IT

• High processing failures in the real-time data flow pipeline. Consumed Apache Kafka and RabbitMQ using Java and SQL Server for real-time processing. Decreased processing failures by ~65% in FY22 Q4.
• Client-server web application performance was lagging. Optimized performance utilizing lazy-loading and SOLID design patterns in Java and JavaScript. Reduced TTI to 1.83s, CLS to 0.05s, and LCP to 1.58s.
• The production deployment pipeline had significant risk due to massive unstructured metadata. Translated Python scripts into PostgreSQL and automated scheduling using PCF. Eradicated manual processes and mitigated the risk of deployment errors.
• Unmonitored internal systems caused excessive support failures (>10,000/day). Designed a fault-tolerant alert-logging mechanism using Java, JavaScript, and KQL. Provided real-time observability across multiple KPIs.
• Internal and external tools were untested for high traffic. Conducted performance testing using JMeter, Prometheus, and Grafana. Achieved a sustainable load of 1M DAU.

High Radius

2019 - 2020

Software Engineer Trainee

• The business needed predictive tooling for client payment behaviors. Developed an AI-based Fintech web app using Python, Flask, AWS, and JavaScript. Successfully delivered a supervised regression model predicting B2B payment dates reliably.

Ericsson

2019

Trainee (Apprenticeship)

• Application deployments were slow and highly manual. Developed a modern CI/CD pipeline using AWS, Kubernetes, and Docker. Streamlined deployment processes and reduced overall release time by 70%.

Aakar Software & Services

2018 - 2019

Software Engineer Intern

• The monolithic React web application suffered from extremely slow load times. Leveraged lazy-loading concepts across the frontend architecture. Decreased the apparent load time of the application to near-zero (0.001ms).
• The weekly newsletter delivery was a highly manual workflow prone to delay. Automated the pipeline using Apache Airflow. Reduced manual effort by 90% and ensured 100% on-time delivery.