Machine Learning Software Engineering

NITHESH VEERAPPA

313-***-**** Farmington Hills, MI - 48334

********@*****.*** linkedin.com/in/nitheshveer github.com/nitheshkv96 PROFESSIONAL SUMMARY

ML-focused Software Engineer with 6+ years delivering production-grade real-time AI systems for computer vision, sensor fusion, and embedded/cloud deployments and embedded control software for Automotive application. Skilled in scalable ML pipelines, low-latency inference, distributed microservices, and Agile development. Proven track record of reducing inference latency, boosting detection accuracy, and automating multi-terabyte workflows. SKILLS

Technical Skills Machine Learning, Computer Vision, Natural Language Processing, Deep Learning, Reinforcement Learning, Data Structures & Algorithms, OOP, Design Patterns Programming Python, Matlab, C++, Go, Cuda Programming, SQL ML/DL Tools PyTorch, Sk-learn, PySpark, OpenCV, Pandas, Transformers, OpenAI, Langchain Distributed Systems gRPC, WebSockets, Kafka, Docker, AWS(S3, Sagemaker, IAM, Lambda, SNS, EC2), ETL Tools Matlab and Simulink, EA, JIRA, GIT, SVN, DOORS, Docker, ROS2 Software Engineering Agile, V-cycle, Microservices, CI/CD, Docker, Git, MLOps, Streaming Data Pipelines IDE VSCode, Vim, Jupyter Notebook, Google Colab

Operating Systems Windows, Linux

EXPERIENCE

AI Algorithm Developer — STONERIDGE INC, Michigan, USA June 2024 - Present

• Architected Lidar-Camera processing pipeline including sensor-fusion, cross calibration and algorithms for object detection and tracking improving depth estimation performance.

• Developed server-client based application for data collection using websocket programs and ROS2 with playback capability facilitating algo developemnt(ISP, ADAS features).

• Designed a targetless real-time calibration service (Kinematics + particle filters) running concurrently with ADAS features facilitating the calibration of cameras and Lidars using gRPC achieving remote communication between machines.

• Built cloud-connected ETL pipeline (AWS S3, EC2) for multi-terabyte ADAS datasets, enabling faster analytics.

• Integrated deep learning models into embedded systems for real-time object detection, leveraging the ONNX platform and CUDA-optimized pre and postprocessing pipelines.

• Designed a hybrid unsupervised-supervised ML architecture combining LSTM and MLP networks (zero-shot learning) for torque estimation and trailer presence detection using vehicle CAN data.

• Developed health monitoring service in Golang for CPU/GPU performance and data storage capacity during concurrent streaming from 8 cameras + 2 LiDARs, detecting overloads. AI Algorithm Intern — STONERIDGE INC, Michigan, USA May 2023 - May 2024

• Engineered YOLOv3-based real-time detection and CV modules (DBSCAN, RANSAC) for camera monitoring system.

• Developed Deep Variational Autoencoder for trailer angle estimation; reduced model size by 40% via knowledge distillation for edge deployment.

• Created hybrid ML pipeline (SVD + Decision Tree) for trailer detection, improving inference speed by 100 over baseline.

• Automated post-training evaluation, hyperparameter tuning, Regression testing and A/B testing pipelines, shortening experimental cycles.

Senior Developer — TATA ELXSI, Trivandrum, India Jul 2018 - Jul 2022

• Led the development of production-grade ADAS and EV control software with strict compliance to ISO-26262, ASPICE, AUTOSAR.

• Implemented ML-enhanced model predictive control for MIMO systems, improving accuracy.

• Built Python/MATLAB automation tools for testing and performance profiling, cutting verification cycles by 50%.

• Managed 4-engineer team delivering optimized control software with faster runtime in production. EDUCATION

MS in Artificial Intelligence, University of Michigan, USA Aug 2022 - May 2024 GPA: 4.0/4.0

Coursework: Artificial Intelligence, Natural Language Processing, Text Mining & Information Retrieval, Deep Learning, Com- putational Learning, Pattern Recognition & Neural Networks, Optimization, Stats and Probs BE in Automobile Engineering, Visvesvaraya Technological University, India Aug 2013 - Aug 2017 CGPA: 8.5/10.0

PERSONAL PROJECTS

Automated Algo Trading Service (Go, Python): Designed and implemented a high-concurrency, Go-based back Test- ing Engine and automated trading platform using Interactive Broker API supporting real-time order execution and portfolio management. Deployed a novel Statiscal arbitrage algorithm into an active trading pipeline. Product Management Agentic AI: Developed an LLM-powered agent incorporating Retrieval-Augmented Generation (RAG) and fine-tuning (PEFT, LoRA) capabilities to assist in product management tasks, enabling dynamic information retrieval and contextual decision-making.

FineTuning Gemma LLM with reasoning data: Applied Quantized Low Rank Adaptation (QLoRA) technique to fine tune Gemma(2B) LLM on custom reasoning data for agentic behaviour AI Pair-Programming Assistant Developed an application for assisting the software developer by tapping into the coding capabilities of a pretrained Large Language Model(LLM) using pre-engineered specialized prompts for several code refactoring use cases such as code improvisation, completion, debugging etc. Image Caption Generator Built a hybrid neural network model combining a pretrained Inception-V3 model for image feature extraction and an LSTM network with GloVe embeddings for text sequence learning to generate natural sentences describing an image.

INTERESTS

• Edge AI and On-Device ML

• High-Performance Computing

• Generative AI and LLM Agents

• Algo Trading Systems and FinTech

• Distributed Systems

Contact this candidate