NIKHIL RAJ
I don't just write code. I architect real-world systems bridging hardware, intelligence, and secure infrastructure. Building the operating systems for tomorrow's machines.
System Architect Profile
Nikhil Raj
Systems Engineer
# Engineering Profile loaded...
I operate at the intersection of artificial intelligence, hardware systems, and secure network architecture.
While many build isolated algorithms, my focus is on integration: deploying neural networks onto edge devices, bridging sensors with logic, and ensuring the resulting automated systems are resilient against cyber threats.
The future isn't just software. The future is code that moves. I build the systems that power it.
# EOF ▮
Deployed Systems
Rakshak AI v3.2
An autonomous threat intelligence system utilizing edge-deployed neural networks for real-time spatial analysis and behavioral anomaly detection. Designed for defensive robotics applications.
secure-connect-chat
Secure real-time chat application built with React, Vite, and Supabase. Features a robust frontend utilizing shadcn-ui and Tailwind CSS for optimal user experience.
Branch IQ
Scalable intelligence API handling multi-threaded data ingestion and real-time semantic processing. Built to withstand high-volume concurrent requests with sub-100ms latency.
Medic-pro
Deterministic medical interaction engine. Uses strict algorithmic verification to cross-reference pharmaceutical combinations, preventing critical chemical conflicts.
SHIELD-NET
Post-quantum cryptography based secure communication channel established for the defense sector to communicate. Features mesh topology for robust message packet routing.
Engineering Vision
Autonomous Infrastructure
We are moving past human-in-the-loop systems. My focus is building resilient, self-healing networks that can interpret physical sensor data and execute logic without manual oversight. Code that acts in the physical world.
Zero-Trust Security
When machines make decisions, security is no longer an add-on; it's the foundation. I engineer from a zero-trust perspective, ensuring every node, API call, and hardware handshake is cryptographically verified.
Hardware-Software Synthesis
Software is limited by the glass screen. Robotics breaks that barrier. I am driven by the synthesis of low-level motor control with high-level cognitive models—building the connective tissue between thinking and moving.
System Timeline
secure-connect-chat
Developed a secure real-time chat application with end-to-end encryption capabilities.
SHIELD-NET
Architected a post-quantum cryptographic mesh network for defense sector communication.
Rakshak AI
Developed edge-AI computer vision and hardware integration for defensive robotics applications.
Medic-pro
Building deterministic medical interaction engine utilizing strict algorithmic verification.
Branch IQ
Planned deployment of scalable intelligence API handling multi-threaded data ingestion.
Operational Domains
System Failures & Lessons
A true engineer's credibility isn't built on successes alone. It's built on the wreckage of failed prototypes and the systems rebuilt from them.
Rakshak GPS Drift Anomalies
The Breakdown: Initial hardware tests relied on consumer-grade GPS for outdoor positioning. Drift caused the unit to register false-positives and attempt navigation through solid obstacles, leading to physical collision.
The Fix: Scrapped pure GPS. Implemented a sensor fusion algorithm utilizing IMU data, wheel odometry, and LIDAR SLAM to create a robust local coordinate frame resilient to GPS multi-path errors.
BranchIQ Monolith Crash
The Breakdown: Attempted to run synchronous voice-processing logic on the main application thread. During a stress test, a spike in concurrent requests caused thread exhaustion, bringing down the entire DB connection pool.
The Fix: Decoupled the architecture. Moved heavy inference tasks to a dedicated asynchronous worker queue (Redis/Celery), keeping the main API gateway responsive even under heavy load.
DoseWiz False Negatives
The Breakdown: The early interaction model achieved 99% training accuracy but failed spectacularly on unseen clinical data, failing to flag obscure but critical drug interactions.
The Fix: Realized statistical ML was the wrong tool for deterministic medical logic. Rebuilt the core engine using a rules-based expert system paired with an ontology graph, ensuring 100% predictable outcomes.