Devesh Raj

Senior Software Engineer

Bengaluru, Karnataka, India0 mo experience

AI EnabledAI ML Practitioner

Key Highlights

Developed AI framework for 270 million devices.
Optimized performance and memory usage in AI systems.
Increased client transactions through targeted marketing.

Stackforce AI infers this person is a Backend-heavy AI Engineer with strong capabilities in machine learning and data analysis.

Contact

Skills

Core Skills

AiMachine LearningData Analysis

Other Skills

AOSPBashC++Convolutional Neural Networks (CNN)Deep LearningDjangoGenerative AIGitHALLarge Language Models (LLM)PerforcePyTorchPython (Programming Language)SQL

Experience

Samsung r&d institute india

2 roles

Senior Software Engineer

Jul 2022 – Present · 3 yrs 8 mos

Developed and commercialized a large-scale AI model deployment framework used across 270 million devices, creating an efficient, coordinated ecosystem for specialized AI services. Drove core performance, stability, and efficiency improvements:
Engineered a native C++ system pipeline enabling 50+ developer teams to implement DNN/Gen AI models independently of hardware specs.
Optimized the AOSP HAL server performance by 15% and implemented an efficient shared memory data layer, reducing memory usage by 10%.
Guaranteed execution integrity and fault tolerance by resolving 100% of race conditions and implementing two-way death notification, cutting issue resolution time by 20%.

C++AOSPHALAIMachine LearningDeep Learning

Student Trainee

Jan 2022 – Jul 2022 · 6 mos

Achieved a 7% optimization in memory allocation for a cutting-edge OnDevice AI framework interface. This was accomplished by leveraging Android's Dmabuf utility to successfully eliminate redundant data copies.

Capillary technologies

Intern

May 2019 – Jul 2019 · 2 mos · Bengaluru Area, India

Drove a 5% increase in repeat transactions for key clients (e.g., Peter England, Allen Solly) by scheduling 100+ targeted marketing campaigns. Leveraged SQL for KPI data analysis to perform root cause analysis and address sales declines across various brands.

SQLData Analysis