Portfolio

About Me

Back-End Engineer & AI/ML Enthusiast | BTech CSE (AI/ML) at Christ University

Currently working as a Back-End Engineer Intern at GoQuant, where I architect multi-service systems using FastAPI, REST APIs, and WebSocket integrations. I specialize in building high-performance systems with Python and C++, working extensively with PostgreSQL, Redis caching, and low-latency market data pipelines that achieve 99.9%+ accuracy.

Previously at XenReality Technologies, I enhanced OCR accuracy by 15-20%, trained object detection models with 85% precision, and achieved 85% accuracy in real-time pose estimation while reducing GPU memory consumption by 25% through optimization techniques.

With expertise spanning Computer Vision, Deep Learning, and systems programming, I'm passionate about building scalable, efficient solutions. I'm also a game developer with strong skills in Unreal Engine, C++, and GPU programming, constantly pushing the boundaries of what's possible in both backend systems and interactive experiences.

Work Experience

Back-End Engineer Intern

GoQuant

Remote

June 2025 - Jan 2026

Architected and integrated DEX exchange into existing CEX infrastructure under a tight timeline, resulting in $250K+ trading volume from a new client while leading a team of 5 engineers, creating a reusable integration framework that accelerated onboarding of 4 additional DEX exchanges.
Designed a smart order routing mechanism that selects the best bid from a consolidated order book across multiple exchanges, improving average execution price by 18%.
Built low-latency market data pipeline achieving 99.9% accuracy on high-volume streams using C++ multithreading and Kafka integration with Order Execution service.
Reduced order placement latency by 40% and eliminated internal execution failures across 10+ exchanges by architecting a unified order management system with optimized WebSocket handlers and REST APIs in Python and C++.

AI & ML Software Engineering Intern

XenReality Technologies

Hybrid

August 2024 - May 2025

Achieved 10x production throughput increase replacing manual inspection with real-time automated anomaly detection (95.5% image & 97.3% pixel level accuracy), eliminating quality control bottlenecks for manufacturing client.
Designed and deployed an instance detection model with 80% zero-shot and 92% one-shot accuracy on medical equipment, enabling rapid inventory tracking.
Implemented real-time pose estimation with 85% accuracy in tracking body movements and coordinates, enabling advanced motion analysis resulting in 37% productivity increase for client.
Improved OCR text recognition accuracy by 15-20% through advanced pre-processing techniques and algorithms, enhancing document processing capabilities and data extraction efficiency.