ML Engineer · AI Engineer · Cloud Architect

Nagizaaz Shaik

It doesn't matter where you start. It matters that you build.

Intern at Uber writing anomaly scripts. Now: production ML for 450K members, zero audit findings, five models in deployment, clinical AI shipping FHIR R4 to Epic. The journey doesn't stop here.

◈ ML models · AI agents · Cloud infrastructure · Tech & Healthcare
Connect on LinkedIn View GitHub Resume ↓
Scroll
Experience

Where I've shipped.

Jul 2024 – Jun 2026
Current
Cloud Architect
Wright-Patt Credit Union · Fairborn, OH
  • Designed and owned the full ML platform on AWS — 5 production SageMaker endpoints, MLflow model registry with version-gated promotion. Deployment failures reduced 85% in 60 days.
  • Built SageMaker Model Monitor with custom statistical baselines — caught 2 critical model degradation events before accuracy dropped in production credit decisions.
  • Deployed XGBoost underwriting model processing 500+ automated loan decisions/day. Cut manual analyst review queue 30%.
  • Fine-tuned BERT on internal ticket data as a FastAPI microservice — sub-200ms p99, 40% reduction in misrouting across 10K+ weekly requests.
  • Passed 2 consecutive NCUA audits with zero findings on $400M in member financial data — AES-256, IAM least privilege, CloudTrail on all inference calls.
Jul 2020 – Jun 2023
Data Engineer
Cognizant · Hyderabad, India
  • Led Oracle 11g → Amazon Redshift migration — redesigned schema for columnar storage, rewrote 20+ stored procedures as Airflow DAGs. Query runtime improved 60% on 25GB+ warehouse.
  • Designed 15 production ETL/ELT pipelines in Airflow across Oracle, SQL Server, flat files, and REST APIs — surfaced a silent corruption issue invalidating 3 months of downstream financial reports.
  • Built real-time Kafka feature pipeline for retail demand forecasting — POS events in 5-min tumbling window, end-to-end latency under 60s, contributed to 28% reduction in emergency restocking.
  • Promoted to tech lead in 2 years — code reviews for 3 engineers, 100% on-time delivery on 2 client accounts.
Jan 2019 – Mar 2019
SQL / Data Intern
Uber · Hyderabad, India
  • Return engagement on the same city ops team — built SQL queries and Vertica views for weekly driver utilisation, trip completion, and surge pricing reporting across Hyderabad.
  • Consolidated fragmented ad-hoc queries into standardised templates, cutting analyst prep time for the weekly ops review from 3 hours to under 45 minutes.
Aug 2018 – Oct 2018
Python Intern
Uber · Hyderabad, India
  • Automated extraction and transformation of operational event data using Python and pandas — converted a 4-hour manual weekly process into a scheduled job producing clean summary tables for city ops.
  • Built an anomaly detection script flagging unusual trip counts and revenue figures against rolling 7-day baselines before data entered the reporting pipeline.
  • Secured a return internship offer by approaching the manager directly at the end of the engagement.
42
GitHub Repos
4+
Years Experience
9
Public Projects
5
Cloud Certifications
Projects

Things I've actually built.

9 open-source repos. Real code, real tests, real CI. AI agents · clinical AI · privacy ML · data infrastructure.

AI Agents ↗
Agentic Pipeline Healer
LangGraph multi-DAG orchestrator that monitors Airflow pipelines, classifies failures, generates AST-level code fixes, and dispatches Slack Block Kit alerts with a full SQLite audit log.
LangGraphAirflowASTFastAPI
● Live repo CI passing
Computer Vision ↗
CheXpert Pathology Classifier
DenseNet121 + EfficientNet-B4 ensemble on the CheXpert 14-label benchmark. MC Dropout uncertainty, Grad-CAM + ScoreCAM explainability, DICOM pipeline, and clinical PDF report generator.
PyTorchDenseNet121Grad-CAMpydicom
● Live repo 92% AUC
LLM Safety ↗
LLM Safety Auditor
Automated red-teaming framework generating 250+ adversarial attacks across 6 mutation strategies. Scores against OWASP LLM Top 10 and NIST AI RMF. Produces PDF audit reports with risk matrices.
sentence-transformersReportLabOWASP
● Live repo 250+ attack vectors
Clinical ML ↗
Clinical Survival Analysis
6-model benchmark: Kaplan-Meier, Cox PH, Random Survival Forest, XGBoost Survival, DeepSurv, and DeepHit. Competing risks with Aalen-Johansen CIF. Validated on NCCTG, GBSG2, and synthetic ICU cohorts.
lifelinesscikit-survivalPyTorch
● Live repo 6 model types
Privacy ML ↗
Federated Credit Risk
3-institution federated learning via Flower FedAvg — zero raw data sharing. L2-norm model poisoning guard, gradient clipping, ECOA Fair Lending feature audit, and differential privacy integration.
FlowerPyTorchFastAPIECOA
● Live repo Zero data sharing
Privacy AI ↗
Differential Privacy LLM
DP-SGD fine-tuning from first principles with Rényi differential privacy budget accounting. Interactive Streamlit dashboard visualizing privacy-utility tradeoffs across ε values and noise multiplier configurations.
PyTorchOpacusFastAPIStreamlit
● Live repo Rényi DP accounting
Data Engineering ↗
Kafka Stream Feature Store
Real-time ML feature store: Kafka producers → Redis with sub-60s freshness guarantee. JSON serialization, per-IP rate limiting, and a FastAPI serving layer with feature versioning and TTL management.
KafkaRedisFastAPIPydantic
● Live repo < 60s freshness
MLOps ↗
SQL-to-DAG Compiler
Transpiles Oracle SQL/PL-SQL and dbt models into production Airflow 2.x DAGs. Handles CTEs, subqueries, stored procedures, and edge cases. Exports lineage graphs in Mermaid and DOT formats.
sqlparseJinja2AirflowNetworkX
● Live repo Lineage export
Computer Vision ↗
Sign Language Translator
Real-time ASL recognition using MediaPipe 21-point hand landmarks fed into a custom PyTorch CNN. 96.7% test accuracy with gesture smoothing via majority-vote buffer. 75 tests, 93% coverage.
MediaPipePyTorchOpenCVscikit-learn
● Live repo 96.7% accuracy
Skills

Full ML lifecycle fluency.

From raw data to deployed, monitored model — across cloud platforms.

LLMs & AI Agents
LangChain LangGraph RAG FAISS Prompt Engineering OWASP LLM Top 10 Hallucination Detection Red-teaming
Machine Learning
PyTorch HuggingFace scikit-learn XGBoost SHAP / LIME Opacus (DP-SGD) Flower (FL) Grad-CAM
MLOps & Infrastructure
MLflow Airflow SageMaker Evidently AI Docker GitHub Actions FastAPI Terraform
Data Engineering
Apache Kafka Apache Spark dbt Snowflake Redis PostgreSQL Redshift SQL Lineage
Cloud & Languages
AWS GCP Azure Python SQL / PL-SQL Bash Scala
Security & Governance
HIPAA GDPR ECOA / Fair Lending PII Redaction Differential Privacy Audit Logging FHIR R4
Education & Certifications

Credentials.

Education
Doctor of Business Administration (DBA)
Belhaven University · Jackson, MS
Aug 2026 – Present · Business Intelligence & Analytics
M.S. Computer Science
University of Dayton · Dayton, OH
Aug 2023 – May 2025 · GPA 3.93 / 4.0
B.Tech Computer Science Engineering
CMR Engineering College · Hyderabad, India
Aug 2016 – May 2020 · GPA 8.9 / 10.0
Certifications
AWS
AWS Certified Machine Learning – Specialty
AWS
AWS Certified Cloud Practitioner
GCP
GCP Professional Data Engineer
❄
Snowflake SnowPro Advanced: Data Engineer
dbt
dbt Analytics Engineering Certification
Contact

Let's build something.

ML models · AI agents · cloud infrastructure.

in LinkedIn ⌥ GitHub ↓ Resume

Verify you're human to unlock contact