CASE NO. 43 — FINDINGS PUBLISHED
Three years in: Ajitesh Sharma builds high-throughput DPI engines, and deploys streaming anomaly detectors.
Currently completing his B.Tech in Computer Science (Information Security) at VIT Vellore.
He interned at Hindalco Industries (Aditya Birla Group), digitizing workforce scheduling for 1,200+ site personnel.
He builds things that inspect, detect, and protect. His work spans C++17 deep packet inspection engines, O(1)-memory streaming anomaly detectors, and containerized microservices platforms.
When he's not writing lock-free code, he's researching or filing Indian patent applications for systems that don't exist yet, restless lad.
A high-performance C++17 Deep Packet Inspection (DPI) engine utilizing JA4+ fingerprints and Shannon Entropy for real-time encrypted malware traffic detection. Engineered for multi-threaded packet capture and low-latency feature extraction.
O(1)-Memory anomaly detection pipeline using LightGBM and Fast-Hoeffding Trees, deployed via FastAPI for real-time inference with TreeSHAP explanations.
Containerized seat reservation microservices with high-availability scheduling. Built with robust CI/CD and system observability.
AI-powered dietary analysis tool leveraging Gemini Vision API for food recognition and nutritional breakdown.
Enterprise system for digitizing workforce scheduling, reducing manual allocation overhead for 1,200+ site personnel daily.
C++17, Java, Python, C, SQL, PL/SQL, JavaScript.
Data Structures & Algorithms, OOP, OS, Computer Networks.
DPI, JA4+, YARA, Wireshark, eBPF, Linux, Autopsy, Metasploit.
Spring Boot, FastAPI, Docker, Kubernetes, Jenkins, AWS, PostgreSQL, REST, WebSockets.
PyTorch, LightGBM, Scikit-Learn, TreeSHAP.
Architected and developed a digitized workforce scheduling system, migrating from manual legacy processes. Implemented robust Role-Based Access Control (RBAC) ensuring strict data segregation across administrative tiers. Streamlined the daily shift allocation and attendance monitoring for over 1,200 site personnel.
Pursuing rigorous coursework in Operating Systems, Network Security, Cryptography, and Advanced Data Structures. Core member of ACM-VIT.
This paper investigates the unforeseen consequences of over-aligning proactive AI agents to maximize safety at the expense of social utility.
READ FULL ANALYSIS →System and Method for O(1) Memory Streaming Anomaly Detection
Audio Deepfake Detection via Shannon Entropy
Real-time Lip-Sync Detection using Visual CNNs