About me
Careful analysis. Clear conclusions.
Hi—I'm Nellie.
Applied ML and data analysis with an engineering mindset.
I'm an applied machine learning practitioner with a strong quantitative and software engineering foundation, currently completing an M.S. in Computer Science (Machine Learning). I focus on building ML models and systems that are useful, measurable, and grounded in real-world constraints.
My recent work centers on applied ML projects that combine vision, NLP, and retrieval—multimodal pipelines, RAG systems, and evaluation-driven experiments. I care about problem framing, establishing baselines, running controlled comparisons, and understanding why a model behaves the way it does, not just whether it runs.
Through graduate coursework and hands-on projects, I've worked end-to-end across the ML lifecycle: exploratory analysis, feature design, model training, and evaluation. I'm most comfortable in modeling-heavy roles where metrics, error analysis, and trade-offs matter as much as implementation.
Earlier in my career, I worked as a software engineer on a real-time payments platform, contributing to backend microservices, automated testing, and regulated releases. That experience shaped how I approach ML work today— favoring reproducible workflows, clear interfaces, and systems that are safe to operate and evolve.
I'm targeting applied ML and modeling-heavy data science roles, and I'm also open to ML-adjacent backend teams where experimentation and production live side by side.
ML via graduate study and applied projects
Resume
Professional Summary
Applied machine learning practitioner with a strong software engineering foundation and production experience on a real-time payments platform. Currently completing an M.S. in Computer Science (Machine Learning).
Experienced in end-to-end ML/LLM projects, including modeling, offline evaluation, error analysis, and API-based deployment, with a disciplined approach to testing, CI, and release readiness.
Information
- Florida, USA
- +1 (908) 764-7332
- cordova.nellie@outlook.com
- linkedin.com/in/cordovank
- github.com/cordovank
Technical Skills
- Languages — Python (primary) · Java · SQL · R · C++ (academic)
- Backend & APIs — FastAPI · REST APIs · Pydantic · typed request/response models
- Quality & Delivery — CI/CD (Jenkins, Maven) · automated testing (JUnit/Cucumber) · API/perf testing (Postman, SOAP UI, JMeter) · release readiness
- Cloud & Ops — Docker · AWS (AI Practitioner) · monitoring/alerting · runbooks
- LLM / RAG / Agents — RAG pipelines (retrieval, reranking, citations) · evaluation & guardrails · prompt/tool workflows · OpenAI Agents SDK · LangChain (exploratory)
- ML Foundations — PyTorch · scikit-learn · Transformers · NumPy · pandas · classification · fine-tuning & transfer learning · sequence models (RNN/LSTM, seq2seq) · attention / memory networks
- Data & Evaluation — experiment design · cross-validation · data preprocessing/augmentation · metrics (F1, RMSE, perplexity) · error analysis · visualization (Matplotlib)
- Developer Tools — Git · Jupyter · VS Code · IntelliJ · JIRA · Bitbucket · Confluence
- Other — English & Spanish (bilingual)
Professional Experience
Software Engineer
JPMorgan Chase & Co. | Tampa, FL
2019 - 2021
Real-Time Payments Platform
- Contributed end-to-end across the SDLC for Java-based payments microservices and drove production readiness.
- Authored comprehensive operational runbooks—architecture, dependencies/configs, failure modes, diagnostic checklists, and step-by-step remediation—to enable fast triage, recovery, and self-serve on-call; provided light support on monitoring/alerting dashboards with platform/SRE.
- Built and maintained test automation (JUnit, Cucumber); enforced CI quality gates (tests, static analysis, coverage) to keep merges green, and executed functional/perf testing (Postman, SOAP UI, JMeter).
- Managed regulated releases: compiled release evidence (test/coverage reports, change tickets, approvals) to secure production sign-off; presented services to global production management.
- Led/participated in agile rituals (sprint planning, backlog refinement); authored technical docs; coordinated across time zones.
2019 - 2020
Salesforce Platform
- Delivered Salesforce CRM data-collection and reporting features for a social-impact organization.
- Gathered stakeholder requirements and translated them into schema updates, UI changes, and documented technical implementations.
- Expanded the underlying data model by adding new objects, fields, and relationships, and updated the custom Visualforce UI to align with the revised schema for accurate structured data capture.
- Contributed a small Flow-based automation to digitize an existing paper survey, enabling standardized survey generation within the CRM.
ML Research Assistant
CS Dept. @ William Paterson University | Wayne, NJ
2018 - 2019
- Performed exploratory data analysis on behavioral survey data to complement hypothesis-driven research on teen tanning behaviors.
- Cleaned and prepared structured survey data, handling missing values and categorical variables.
- Built and evaluated predictive and unsupervised models in Python and R, including Logistic Regression, KNN, and clustering methods.
- Compared model behavior and results to understand key drivers, limitations, and trade-offs.
- Communicated findings and insights to faculty and student audiences through presentations.
Education
M.S. in Computer Science (Machine Learning)
Georgia Institute of Technology | Atlanta, GA
2022 - Present
Relevant Coursework: Machine Learning, Deep Learning, Natural Language Processing, Reinforcement Learning, Network Science
B.A. in Mathematics | Minor: Computer Science
William Paterson University | Wayne, NJ
2016 - 2019
Relevant Coursework: Applied Regression Analysis, Data Warehouse & Data Mining, Database Management Systems, Cloud Computing
Magna Cum Laude • Pi Mu Epsilon (National Mathematics Honor Society)
Projects
- All Projects
- AI
- DL
- NLP
- UX
- Backend / Systems
Modular RAG
Flow-based RAG framework — hybrid retrieval (BM25 + FAISS, RRF fusion, MMR reranking),
explicit wiring, vendor isolation, observability-first architecture.
Retrieval evaluation framework included (Hits@k, MRR, NDCG).
Plate2Recipe – Food Image to Recipe Generation
Two-stage multimodal pipeline: ViT ingredient recognition → GPT-2/LSTM recipe generation.
Key finding: lower training loss (100k samples) produced worse outputs than a smaller, better-tuned run (10k samples) — quality ≠ loss.
RAG System with Guardrails
RAG v1 — BM25 + FAISS hybrid retrieval, RRF + MMR, citation-aware answering. Led to Modular RAG architecture.
CRM & Ticketing System
FastAPI backend with Docker, RBAC, and a deployed HuggingFace Spaces demo.
FoodDB
Local-first nutrition facts catalogue — SQLite backend, Typer CLI, and Streamlit UI sharing one service layer. Regex parser + swappable OCR backends (Tesseract / Ollama vision).
ProductLens
LLM-driven product comparison tool that ranks products based on user priorities.
Professional TwinBot AI
AI-powered resume-grounded assistant: tool-style prompting + retrieval patterns for accurate, personalized responses.
Memory-augmented QA System
Attention-based QA over structured knowledge using memory networks.
Notification System Redesign
User-centered redesign of Discord's notification experience.
Let's Connect!
Looking for ML Engineer and AI Engineer roles.
I build RAG pipelines, agentic workflows, and applied ML systems —
with production engineering discipline behind them.
Recruiters: a short note + role link is perfect.
Contact
-
LinkedInlinkedin.com/in/cordovank
-
GitHubgithub.com/cordovank
-