HELLO, I'M
Himanshu
Jack of all trades, master of some.
Seasoned programmer seeking complex challenges.
Quickly masters new languages and technologies, always craving the next big problem to solve. Currently exploring Quantum Computing and Decentralized AI systems.
Featured Projects:
Research
Conflict-Aware Adapter Composition (C-AAC) for Scene Generation
completed
Advanced scene generation using conflict-aware adapter composition techniques. Developed at AuraX (CIE @IIIT-H) with comprehensive benchmark comparisons.
ReNU: Preserving Texture and Color Fidelity in Diffusion Models
ongoing
A proprietary framework designed to address persistent challenges in diffusion-based generative models. ReNU (Regularized Neural Units) integrates enhanced penalty functions into the reverse diffusion process to significantly improve the preservation of fine texture details and ensure faithful color reproduction in generated images. [cite: 4, 5] The system is particularly effective for applications requiring high-fidelity image synthesis, such as virtual try-on.
MoE: Brain-Inspired LLM Architecture
ongoing
A self-directed research project advancing the thesis that the Mixture of Experts (MoE) architecture is a compelling computational analogue to the human brain's principle of functional specialization. The work deconstructs the neuroscientific foundations of brain organization, provides a technical analysis of MoE models, and synthesizes these domains into a novel, brain-inspired hierarchical MoE architecture.
Professional Experience
CTO
- Researching on diffusion models for: realisitc human generation along with VTO (virtual tryon), upscaling so on...
- Building robust backend, deployment strategies, reducing latency, improving performance, and ensuring scalability.
- Using AWS, ShaktiCloud, and other cloud services for deployment and scaling.
Founding Software Engineer - CTO
- Engineered a robust REST API server in microservice architecture, integrating MongoDB, PostgreSQL, and Redis, resulting in a 40% improvement in data retrieval efficiency.
- Successfully optimized the performance of the Web App by reducing loading time by over 80% (200ms), resulting in a smoother and faster user interface.
- Led the system design process, implementing scalable architectures and performing comprehensive unit and integration testing, achieving a system reliability of 99.9%.
- Optimized web application performance, reducing load time by over 80% (from 1s to 200ms), ensuring a seamless user interface and improving retention rates.
AI Researcher
- Utilized DevOps practices with Docker to streamline development and deployment processes
- Implemented code obfuscation techniques to secure code, mitigating risks of code leakage and improving security for public deployment.
- Contributed to the development of TublianOS, optimizing system performance and enhancing user experience
- Completed an intensive masterclass in System Design, gaining insights into designing scalable, resilient systems using best practices and industry standards, applicable to real-world projects, taking google's search engine architecture as an example.
Development Projects
adminNexus
Designed and developed a comprehensive admin dashboard for real-time analytics, logging, and feedback management.
FastSearch
Developed a Rust-based private search engine, providing an alternative to Google Search with privacy-focused features.
Dynamo-Prisma
Developed a tool for dynamically creating Prisma Models from JSON Schema