Harshal Sonawane

I’m a software engineer who focuses on turning complex data and AI problems into reliable, production-ready systems. I enjoy building pipelines that run fast, APIs that stay responsive under load, and models that deliver measurable outcomes.

At GoQuant USA, I integrated real-time market data from Bitget Exchange by fetching L1 and L2 order book data using WebSocket and REST API. I also developed efficient high-frequency trading data pipelines, implemented real-time order book tracking, and improved low-latency processing for algorithmic trading strategies.

At ElevateTrust.ai, I engineered a deepfake detection backend using Python and state-of-the-art CViT (Convolutional Vision Transformer) models on real-time video feeds. I designed and optimized RESTful APIs to serve detection results with scalability and low-latency performance under concurrent user loads, while also coordinating with the frontend team.

My project work reflects the same drive for impact and performance. ForensicVision delivered sketch generation and recognition with 92% accuracy and improved workflow time by 40%; E-Health Management System supported 2,000+ patient records with Razorpay integration and “zero errors,” plus 50% faster MongoDB query response times. I also built a privacy-centric InsiWind tool and optimized performance to minimize flickering.