Sreelakshmi Biju

Sreelakshmi Biju

AI/ML Engineer
UAE
English, Arabic, Malayalam, Hindi
Full-time Hybrid

About Me

AI/ML Engineer with 3+ years of experience in software development, specializing in deep learning, generative AI and LLM applications, and AI on edge and high-performance compute hardware. Strong grounding in computer sy…

Experience

Senior Engineer

Ignitarium Technology Solutions Pvt Ltd · Bengaluru, India
Sep 2023 - Present · 3 years 2 months

<p>Contributed to a graph execution runtime for a custom AI inference accelerator, enabling efficient scheduling and execution of deep learning workloads on specialized hardware.</p><p> Integrated ONNX-based model workflows into the runtime pipeline, translating model operators and tensor dependencies into accelerator-compatible execution graphs.</p><p> Designed kernel-level runtime modules and optimized data structures to represent neural network graphs, operators, and tensor dependencies for optimized traversal and execution.</p><p> Implemented tiling and workload partitioning strategies to map large neural network operations onto constrained on-chip memory, improving memory locality and reducing external bandwidth pressure.</p><p> Optimized data movement and execution pipelines to maximize accelerator utilization and inference throughput across CNN-based and sequential architectures.</p><p> Conducted low-level profiling using hardware traces and instrumentation tools to identify bottlenecks in compute scheduling, memory access, and synchronization.</p><p> Collaborated with hardware architecture, compiler, and ML framework teams to validate model execution and align software with hardware capabilities.</p><p> Designed a deep learning-based semantic segmentation model to classify and segment pixels related to human presence near railway tracks in real-time video.</p><p> Built and trained a computer vision pipeline for pixel-level classification under varying environmental conditions.</p><p> Collected and curated railway track video datasets; performed frame extraction, preprocessing, and augmentation (rotation, scaling, brightness, noise injection) to improve generalization.</p><p> Developed training/evaluation pipelines including dataset preparation, model training, validation, and performance analysis using IoU and pixel accuracy.</p><p> Performed systematic testing, debugging, and validation of preprocessing, training, and inference components.</p>

Skills

Python C++ Java SQL PyTorch TensorFlow Keras Scikit-learn OpenCV ONNX Pandas NumPy LLM-based agents RAG embeddings LangChain Hugging Face Transformers NVIDIA Jetson AGX Orin CUDA OpenCL Azure AWS S3 Docker Git GitHub CI/CD Jenkins SDLC SSDLC PyTest SonarQube PEP8 Visual Studio JIRA Streamlit Anaconda matplotlib PowerBI Tableau Tesseract-OCR NLP Tokenization Machine Learning Semantic Segmentation Data Augmentation Pixel-Accuracy Evaluation Linux Kernel Development Device Drivers Embedded Linux
Report this Profile?