Software Engineer

Soumya Nagendra ****************@*****.*** +1-480-***-****

Portfolio LinkedIn

Education

•

Arizona State University Tempe, AZ

M.S. in Computer Science — GPA: 3.77 Aug. 2024 – May 2026

•

KLE Technological University Hubli, India

B.E. in Computer Science — GPA: 3.65 (9.12/10.0) Aug. 2020 – May 2024 Experience

•

Software Engineering Intern Dec 2025 – Present

Atlas IntelliTek Remote

Trained and deployed a Graph Neural Network (GNN) surrogate model in PyTorch for real-time hydraulic network prediction, achieving 100x inference speedup over physics-based solvers while maintaining accuracy within 3% error.

Built the end-to-end ML pipeline: synthetic data generation from EPANET simulations, graph feature engineering (node/edge attributes from network topology), training with validation splits, and model serving via FastAPI.

Integrated LangGraph-based multi-agent orchestration for CFD simulation workflows, where specialized agents handle preprocessing, execution monitoring, error classification, and auto-retry with parameter adjustments.

Developed evaluation harnesses testing agent behavior across 50+ failure scenarios, measuring recovery rates, false positive rates, and end-to-end turnaround time improvements Projects

•

Directional Stock Price Prediction Using Financial News Dec 2025 ASU – Semantic Web Mining Project

Engineered a multimodal feature pipeline fusing technical indicators (rolling stats, momentum, volume) with transformer- and lexicon-based sentiment (FinBERT, VADER) extracted from 70K+ financial news articles, aligned to 30-minute market intervals.

Built and tuned baseline Logistic Regression and SVM classifiers under purged time-series cross-validation

(eliminating look-ahead bias); the SVM model reached the top F1-score (0.59), and class-imbalance handling

(SMOTE) lifted minority-class recall from 0.28 to 0.70.

•

AWS Cloud-Native Face Recognition Pipeline May 2025 ASU – Cloud Computing Project

Developed an elastic face recognition system on AWS (EC2, S3, SQS) with custom autoscaling (0–15 instances) and a serverless Lambda + ECR pipeline using Dockerized MTCNN and FaceNet, achieving sub-1.2s latency at 100 concurrent requests.

Deployed an IoT edge pipeline via AWS IoT Greengrass v2 and MQTT for on-device face detection, reducing redundant cloud compute by filtering faceless frames at the edge. Programming Skills

• Languages: Python, TypeScript, JavaScript, SQL, C++, Java, Go.

• Frameworks & Library: PyTorch, scikit-learn, Hugging Face Transformers, FastAPI, LangGraph, pandas, NumPy.

• Cloud & MLOps: AWS (EC2, S3, SQS, Lambda, ECR, IoT Greengrass), Docker, MQTT, Git, CI/CD, model serving / deployment.

Research Publications

• N. Soumya et al., “A Fair and Scalable Consensus Algorithm for NFT-Based DAO for Digital Art,” 2023 ICCCNT, Delhi. [DOI]

• N. Soumya et al., “An Efficient Voting-Based Consensus Algorithm for Permissionless Blockchains,” 2023 ICCCNT, Delhi. [DOI]

• N. Soumya et al., “Facial Key Points Detection using MobileNetV2 Architecture,” 2023 I2CT, Pune. [DOI]

Contact this candidate