Machine Learning Engineer | Data Scientist | GenAI & ML Systems Builder
🎓 MS Computer Science @ North Carolina State University

📍 Raleigh, NC
🔗 LinkedIn | 🌐 Portfolio | 📧 ksonawa@ncsu.edu

I’m a Machine Learning & Data Science engineer who enjoys building end-to-end ML and GenAI systems — from messy real-world data and ambiguous business problems to scalable, production-ready solutions.

I’ve worked on customer lifecycle modeling, forecasting pipelines, LLM-powered automation, and RAG systems, combining strong fundamentals in ML, statistics, and systems with hands-on experience in cloud and MLOps.

I especially enjoy problems where data, models, and real business decisions intersect.

My goal is to grow into a full-stack ML / AI engineer who can:

• Design scalable ML & GenAI systems end-to-end
• Build reliable data pipelines and forecasting models
• Translate vague business questions into measurable ML solutions
• Ship models that actually get used in production

Machine Learning & AI

• Supervised & Unsupervised Learning, Feature Engineering
• DNNs, LSTMs, Transformers, LLMs, RAG
• Statistical Modeling, Hypothesis Testing, Forecasting
• Computer Vision (Object Detection, Benchmarking)

Generative AI

• Retrieval-Augmented Generation (RAG)
• Knowledge Graph + LLM systems
• Multi-agent LLM workflows
• Prompt engineering & fine-tuning

Data & Systems

• SQL-based analytics & ETL pipelines
• Time-series forecasting & lifecycle modeling
• Data quality, metric consistency, experimentation

• End-to-end ML pipelines (data → model → deployment)
• GenAI systems using RAG, agents, and knowledge graphs
• Forecasting & lifecycle models for business decision-making
• Computer vision benchmarks and performance trade-off studies
• LLM-powered developer tooling and automation

May 2025 – Aug 2025 | USA

• Modeled customer & product lifecycle signals by engineering 20+ features, enabling stable quarterly renewal & tech-refresh risk estimates
• Reduced prediction variance by up to 9% across 9 customer segments
• Built end-to-end forecasting pipelines (AzureML, AbacusAI) on 3M+ records for renewals and revenue planning
• Designed and owned SQL + Python ETL pipelines ensuring data quality and metric consistency
• Partnered closely with Product & Finance to translate ambiguous business needs into scalable ML solutions

Jul 2023 – Aug 2023 | India

• Identified KPIs and built interactive dashboards using Python, Tableau, and Power BI
• Analyzed engagement trends across domains (Crypto, Covid-19, Spotify vs YouTube)
• Enabled data-driven reporting and performance insights

Aug 2025 – Nov 2025

• Built a 2-agent LLM-based technical debt automation system using GitHub Actions & MCP-enabled VS Code workflows
• Fine-tuned GPT-4o-mini on 140k+ real-world code smells
• Detected 20+ code smells with 70–80% F1-score
• Achieved ~70% safe refactors with end-to-end remediation in seconds (MCP) to minutes (PR-based)

Mar 2025 – Aug 2025

• Developed a biomedical QA assistant using knowledge graphs + retrieval + LLMs
• Processed 10,000+ biomedical entities from structured data
• Implemented a robust fallback mechanism achieving >95% reliable responses

Sep 2024 – Dec 2024

• Designed benchmarking experiments on PASCAL dataset (9,600+ images)
• Optimized models using PyTorch & CUDA
• Quantified speed–accuracy trade-offs:
  • YOLO: 135 FPS @ 75.7% mAP
  • SSD: 30 FPS @ 91.7% precision

Jul 2023 – May 2024

• Built recommendation systems using ML (RF, SVD, PMF) and DL (CNNs, LSTMs, Autoencoders)
• Analyzed Amazon Food Reviews dataset (568K reviews, 256K users)
• Performed SQL-driven feature extraction and statistical analysis for personalization

• NVIDIA — Accelerating End-to-End Data Science Workflows
• SIES GST — DevOps Student Development Program

Girija Welfare Association — Navi Mumbai, India
Jul 2022 – Jul 2024

• Led cross-functional teams for community events and fundraising
• Developed leadership, coordination, and stakeholder management skills

• 💼 LinkedIn: https://linkedin.com/iamkaustubhs
• 🌐 Portfolio: https://kaustubhas.github.io
• 📧 Email: ksonawa@ncsu.edu

⭐ Always open to ML, GenAI, and data-driven engineering opportunities.