Aditya Singh

Spearheaded development of BAM and Realm, SaaS platforms managing over $50M in real estate portfolios for 100+ HNI clients.

Engineered scalable Python/Django backend architecture, boosting data processing speeds by 65% and accommodating 3x user load growth.

Developed an end-to-end automation pipeline for government PDF property detail extraction, leveraging TROCR (Transformers OCR) for captcha solving to achieve a 95% bypass success rate.

Built a comprehensive PDF extraction and OCR system, processing over 1000 documents with 92% accuracy by converting pages to structured text using LLM-based JSON generation via Google Gemini API.

Established efficient PostgreSQL database solutions and caching mechanisms, reducing real-time analytics dashboard query response time by 40%.

Created and deployed React/D3.js data visualization components, boosting client engagement with financial analytics by 35% and improving retention by 22%.

Constructed secure JWT and OAuth2 authentication protocols, eliminating security vulnerabilities and ensuring 100% compliance with financial data regulations.

Orchestrated automated testing and CI/CD pipelines with Jenkins, achieving 99.9% uptime and reducing deployment issues by 60%.

Architected ML-powered real-time depth estimation systems for ADAS applications across 5+ vehicle models, improving inference accuracy by 15% through PyTorch optimization.

Built and deployed production ML pipelines, processing over 8TB of data monthly for computer vision tasks and reducing model training time by 30%.

Led backend service development for Samsung's Code.i LLM platform, serving 250+ internal developers, reducing inference latency by 25% and improving model accuracy by 9%.

Constructed scalable data processing pipelines handling 10+ million entries daily, reducing processing time by 7% and saving approximately 420 compute hours monthly.

Formulated RESTful APIs and microservices for XR applications, impacting 15,000+ users with 40% reduced motion sickness and 28% increased session duration.

Established comprehensive CI/CD pipelines and monitoring for 12+ ML models, ensuring 99.8% service availability across production environments.

Built backend services for time-series analysis solutions, processing over 10 million data points daily and supporting analytics for 3 major product lines.

Structured ETL data pipelines using Pandas and NumPy, accelerating historical data analysis by 45% and reducing memory usage by 32%.