Shivam Kumar

Full Stack Developer (Contract)

Built the NFT Staking App and Raffle App on the FTX-backed Solana Blockchain

Used Rust as the backend language to develop smart contracts required for the app

Used React and Typescript to develop the frontend of the app

Software Engineer I

Worked closely with other data scientists to improve the machine learning model. We later filed a patent as well for the invention we made.

Automated the alignment part of the workflow to create Solar report. I was able to reduce the human workload on that step by 80%. I used computer vision and machine learning for it.

Used OpenCV, PyTorch, and Python extensively in my work

Created the data generation pipeline to create a large imagery dataset required to train machine learning models

AI Engineer

Primarily worked on the NLP technology in machine learning and learned how to manage a large codebase

Designed, developed, and deployed a file management system, like Microsoft OneDrive, for Learngram to manage files of various products

Research Assistant

Devised a novel method to automate the detection of open source vulnerabilities in binary files.

Proposed methods for the efficient extraction of unique features inherent in both source code and binary files.

Deployed machine learning and hierarchical filtering techniques using these unique features to get a prima facie evidence of vulnerabilities. In addition, hashing algorithms and subgraph isomorphism were employed to confirm their presence.

Proposed a novel efficient technique for vulnerability detection using neighborhood aggregated locality sensitive hashing.

Achieved around 60% accuracy in detecting the presence of vulnerable library versions in binary packages. Integrated the algorithm into the engine to achieve over 90% accuracy in the detection of vulnerable libraries used in binary packages.