Nakshatra Singh

Stikeman Elliott LLP is one of Canada's premier business law firms, globally recognized for its expertise in corporate, securities, M&A, banking, tax, and litigation.

At Stikeman Elliott, I architected and deployed an enterprise-grade AI infrastructure on Azure that processed over 20M+ legal documents while maintaining strict compliance with Canadian legal data regulations and firm security protocols. I engineered a multi-layered RAG pipeline incorporating hybrid search with BM25 and semantic search models, implementing advanced metadata-driven reranking algorithms that improved result relevance by 47% while maintaining sub-200ms latency at scale.

A leading employee-owned retail company in the United States, Hy-Vee operates more than 285 supermarkets across eight Midwestern states.

At Hy-Vee, I engineered and deployed a next-generation mobile platform serving over 1.2 million active users with 99.98% uptime across both Android and iOS platforms, implementing a sophisticated micro-frontend architecture that enabled independent deployment of critical features without full app releases. I developed a proprietary Impressions Engine using React Native's Layout Animation API combined with custom native modules that increased ad viewability by 25% through precise viewport tracking and predictive rendering, directly contributing to a 20% revenue uplift in digital advertising.

A software technology company that focuses on providing transformative solutions to our customers in the areas of Enterprise Application Development, Data Warehousing and Business Intelligence, DevOps, Robotic Process Automation, Mobile App Development, Quality Assurance and Manpower Consulting space.

At Technodysis, I spearheaded the development of a mission-critical financial application serving 500,000+ users across 12 countries with 99.95% uptime, architecting a highly resilient React Native codebase with sophisticated error handling, retry mechanisms, and offline-first capabilities that maintained functionality during network disruptions. I engineered a state-of-the-art animation system using Moti and Reanimated 2 that implemented physics-based motion with custom spring configurations, achieving 60fps performance even on low-end devices while reducing JS thread workload by 72%.

VOLT is a B2C lending platform that enables retail customers to get instant, low cost & flexible credit by pledging their MFs & stocks via a 100% digital process.

As the founding React Native engineer at VoltMoney, I architected and built a high-compliance financial platform from ground zero that processed over ₹200 crore in loan disbursements with zero security breaches, implementing a sophisticated monorepo architecture with Lerna that managed 8 interconnected applications with shared codebases while maintaining strict separation of concerns. I engineered a complex loan disbursement workflow orchestrating 6+ microservices via RabbitMQ with idempotency keys, transactional guarantees, and circuit breakers that handled 15,000+ daily transactions with 99.99% reliability during peak loads.

NoCap Meta is a high-functioning Web3 company working on diverse projects spanning the verticals of Metaverse, Tokenization, and Web3 technology services.

As the lead backend architect for a cutting-edge NFT marketplace, I designed and implemented a high-performance, horizontally scalable backend infrastructure capable of handling 10,000+ concurrent users with sub-200ms response times during peak loads, deploying a microservices architecture on AWS ECS with auto-scaling groups that optimized resource utilization while maintaining 99.95% uptime. I engineered a sophisticated NFT minting pipeline integrating Web3.js with Ethereum and Polygon networks, implementing nonce management, gas optimization strategies, and transaction monitoring that reduced failed transactions by 68% while handling 5,000+ daily mint operations.

ProdigalAI is a service-based agency with over 20 years of combined expertise in machine learning, delivering innovative solutions in blockchain, AI-driven crypto risk management, advanced analytics and investment tools, computer vision, edge computing, and cloud analytics.

At Prodigal AI, I engineered and deployed a sophisticated DeFi analytics platform that processed real-time blockchain data from Ethereum, Solana, and Binance Smart Chain networks, implementing a high-performance ETL pipeline with Apache Spark that ingested and processed 2M+ blockchain events daily with sub-5-minute latency. I developed a cutting-edge Uniswap V3 liquidity position analyzer that implemented concentrated liquidity modeling, impermanent loss calculations, and fee accrual simulations, processing complex position data with mathematical precision to provide actionable insights for liquidity providers.

Tvish leverages both 3D models and 2D images to deliver an immersive try-on experience, while also providing intelligent fit recommendations powered by data generated through our try-on features.

At Tvish, I engineered and deployed a groundbreaking AR-powered virtual try-on platform serving 250,000+ users with 99.9% uptime, implementing a sophisticated multi-layer AR architecture with ARCore and ARKit that achieved sub-10ms rendering latency for realistic 3D product visualization. I developed a custom machine learning pipeline with TensorFlow Lite that processed real-time camera feeds to detect body landmarks with 94% accuracy, enabling precise virtual garment placement that reduced sizing errors by 62%.

Devtown is a research- and education-driven startup that empowers students to explore and build projects in AI, ML, and Web Development.

At Devtown, I architected and deployed a cutting-edge machine learning education platform serving 200+ students with personalized learning experiences, implementing a sophisticated recommendation system that analyzed student progress and learning patterns to deliver customized content with 89% relevance accuracy. I engineered a state-of-the-art neural machine translation system using Transformer architecture with multi-head attention mechanisms, achieving BLEU scores of 38.7 on WMT'19 English-French translation tasks through advanced tokenization strategies and beam search optimization.