About Me
Hey there! I'm Prasant, a Computer Science student who likes to work in large scale infra and systems engineering. Currently a Meta PE Fellow and former RA working with ML models, I'm also really interested in research in general.
I've recently built Lavender, an open source distributed Linux EDR system with a Rust eBPF kernel module and Go-based detection workers.
I LOVE doing hackathons and working with large scale infra, so hit me up on linkedin if you wanna team up for hackathons or any project really!!
Experience
Production Engineering (SRE) Fellow
Meta, Remote, San Francisco, CA
June 2026
- •Selected as one of ~50 Production Engineering Fellows (SRE) to work on Meta's systems infrastructure — contributing to reliability, observability, and automated deployment of production services.
AI/ML Research Assistant
Cyber Innovation Lab | USM, Hattiesburg, Mississippi
Sept. 2025 - Feb. 2026
- •Built and trained a ModernBERT-based co-attention neural network in PyTorch for social engineering email detection, achieving 100% accuracy on a held-out test set of 158 email pairs.
- •Engineered an email pair processing pipeline from a preprocessed dataset derived from 500,000+ corporate emails, generating 171 structured emails across 4 attacker profiles.
- •Trained a 151M+ parameter transformer with AdamW over 20 epochs, reducing validation loss from 0.57 to 0.0009 while maintaining perfect precision, recall, F1, and AUROC
- •Integrated the model with an upstream GNN-based anomaly detector, achieving 92% end-to-end system accuracy for social engineering detection
Software Engineering Intern
Optimal Answers, Gulfport, Mississippi
Jun. 2025 - Aug. 2025
- •Built backend features for internal and customer-facing web applications using C#, SQL, and REST APIs, supporting day-to-day workflows for business and operations teams.
- •Refactored performance-critical backend logic and SQL database queries, reducing UI latency and cutting employee task time by 30% across core application flows
- •Collaborated closely with design, QA, and data teams to deliver 5+ production features, participating in requirement discussions, testing, and rollout.
Projects
Lavender — Open Source EDR
Distributed Linux endpoint detection and response system with real-time threat visibility.
- •Built a distributed Linux threat detection system in Go with real-time dashboard, detection workers, and structured event logging for hardware and system diagnostics.
- •Wrote a Rust eBPF kernel module intercepting Linux syscalls and kernel tracepoints; structured event logs for anomaly detection and hardware diagnostics.
- •Implemented NATS-based event streaming as a single source of truth, ensuring ordered, lossless data flow across detection workers and dashboard.
Mang0S-x86 Operating System
Open Source x86 operating system built with 5 other devs
- •Implemented a custom x86 Operating System bootstrapped via the Limine bootloader, transitioning from bootloader state into protected kernel execution
- •Built a kernel-level framebuffer terminal with direct video memory access, text rendering, scrolling, and cursor control without libc.
- •Designed low-level memory and I/O abstractions to safely interface with hardware from kernel space.
No-Brainrot - Productivity Focused Browser Extension
Browser Extension that helps you stay focused by blocking short-form content and tracking your web time.
- •Short-form content blocking Chrome browser extension across multiple platforms with 289 installs & 20 daily users.
- •Built a browser extension that blocks certain sites and tracks time, using TS, Webpack, IndexedDB, and Chrome Extension APIs.
- •Eliminated a race condition in IndexedDB writes by serializing add/update operations with a queue and transaction guards, ensuring consistent time-tracking data.
- •Structured background, content, popup, and dashboard modules with shared types and utilities, improving code reuse and maintainability across extension surfaces.
- •Added time analytics for the dashboard to surface usage insights while keeping computation local for privacy.
Personal Status- App made using C++
C++ desktop widget for live portfolio status , sharing your thoughts & active apps in real-time.
- •Built real-time desktop monitoring system using C++ and Windows API that tracks development activity and pushes status updates to portfolio website via secure REST API
- •Developed cross-platform web integration with threaded architecture, local HTTP server, and Vercel API endpoint featuring API key authentication and JSON data exchange
- •Publically released the app, used by a few people!
Timeonar- AI powered Research Helper
A full-stack web app that converts Research Queries into interactive research timeline.
- •Built a full-stack web application that converts natural-language queries into interactive literature timelines using a backend LLM/AI API orchestration service.
- •Made a React + TypeScript frontend and a .NET C# REST API to deliver validated requests and structured, citation-backed responses, improving reliability of aggregated results.
- •Designed backend workflows implementing multi-stage pipelines with failure isolation, partial result propagation, and incremental response streaming, reducing perceived latency.
Chell- A Shell Made In C
An Unix like Shell Made using C programming language compatible with both Linux and Windows.
- •An Unix like Shell Made using C programming language compatible with both Linux and Windows.
- •Implemented core shell functionality including command parsing and execution
- •Cross-platform compatibility with memory management optimization
Education
Bachelor of Computer Science
University Of Southern Mississippi, Hattiesburg, Mississippi
August 2024 - Present