◇01
ML Platforms & Data Quality
Centralized platforms that turn ad-hoc detection into a reliable, governed capability across an org.
- Anomaly detection at warehouse scale
- Continuous data-quality monitoring
- InnerSource governance models (PMC)
Available for ML strategy & advisory
AnupamPanwar.
Machine Learning & Engineering at Apple. I design and lead ML platforms — anomaly detection, data quality, search ranking — that quietly run in production at the scale of hundreds of millions of users.
San Francisco Bay Area · 13+ years · Senior Member IEEE
Shipped at
01 / About
A practitioner of ML at scale.
For 13+ years I've sat on the seam between software engineering and applied machine learning — from intrusion-detection research, to recommendation systems at Yahoo Finance, to learning-to-rank at Walmart, to platform-scale anomaly detection at Apple.
Today I lead the technical roadmap for a centralized anomaly-detection and data-quality ML platform now used by 250+ engineers across half a dozen Apple divisions. The work I'm proudest of is the quiet kind — the platform that prevents a launch from blowing up, the privacy scanner that keeps compliance continuous, the ranker whose effects only show up in revenue.
I mentor early-career engineers, speak at industry events (most recently the Databricks Data+AI Summit 2025), and am a Senior Member of IEEE.
250+
Engineers on platforms I've built
100M+
Emails / day classified in production
1M+
Daily users served by recommenders
9-figure
Revenue impact at Walmart Search
Stack
- Python
- Java
- Spark / PySpark
- TensorFlow
- Scikit-learn
- Hadoop / Hive
- MLlib
- Deep Generative Models
- Learning-to-Rank
- Anomaly Detection
- NLP
- MySQL
- AWS S3
- Jenkins
Education
Stanford University
Artificial Intelligence Professional Program
2025 — 2026
Arizona State University
M.S., Computer Science
GPA 4.0 / 4.0 · First in class
2015 — 2017
Indian Institute of Technology, Delhi
B.Tech + M.Tech (5 yr), Mathematics & Computer Science
2008 — 2013
02 / Expertise
Where I do my best work.
▲02
Search & Recommendations
Learning-to-rank, semantic matching, and recommender systems that move revenue and engagement.
- Founding ML engineer for Walmart.com reranking
- Implicit-feedback recommenders at Yahoo Finance
- Embedding-based query/item matching
✦03
Privacy, Security & Trust
Detection systems that make compliance and security continuous instead of audit-driven.
- Enterprise PII detection across 100% of columns
- Anti-fraud architecture & cross-functional alignment
- Threat-intelligence fusion (research + production)
◐04
Applied AI Research
Bridging research and production — explainability, reasoning-based detection, and peer-reviewed work.
- Reasoning-based anomaly detection (preprint)
- Explainability for global media models
- 5+ peer-reviewed publications
03 / Selected Work
Case studies from a decade of shipping ML.
Apple
Case studyAnomaly Detection & Data Quality Platform
Centralized ML platform for detecting anomalies and data quality issues across Apple's data warehouse — the kind of platform that prevents a launch from blowing up.
- Scaled adoption from 1 → 250+ users across 6+ engineering divisions
- Intercepted a 5× anomalous spike in cloud requests prior to a major OS release
- Founded an InnerSource PMC governing distributed contributions
- Anomaly Detection
- Platform
- Python
- Spark
Apple
Case studyEnterprise PII Detection System
Two-year cross-functional initiative to make privacy compliance continuous instead of audit-driven.
- Automated PII scanning across 100% of data warehouse columns
- Detection algorithms integrated into the core data platform
- Analyzes column metadata + sampled data at warehouse scale
- Privacy
- Compliance
- Data Platform
Walmart Global Tech
Case studyWalmart.com Search Reranking
Founding ML engineer on Walmart.com's reranking team. Built the systems and the science that made search feel relevant.
- Invented the Rerank Micro Service — Walmart's first reranking microservice
- Shipped Learning-to-Rank models with embedding-based semantic matching
- Contributions tied to nine-figure revenue impact
- Learning-to-Rank
- Search
- PySpark
- Jenkins
Yahoo Finance
Case studyStock Recommendation System
Implicit-feedback recommendation system based on page visits — turning behavioral signal into personalized stock picks.
- Recommendations served to 1M+ daily users
- Built on Spark MLlib over Hadoop
- Companion Credibility Score System rated stock analysts 0–100
- Recommendations
- Spark MLlib
- Hadoop
Yahoo Mail
Case studyYahoo Mail Hierarchical Classifier
Productionized a hierarchical classifier mapping coupon and order emails to the Google Product Taxonomy, plus a CNN event classifier and a real-time model-serving service.
- Classifies 100M+ emails per day in production
- CNN-based event classifier for school-event emails
- Designed Resource-as-a-Service (RaaS) for real-time ML model fetching
- NLP
- CNN
- TensorFlow
- Hadoop
Arizona State University
Case studyThreat Intelligence Analytics
Research platform correlating heterogeneous security feeds for intrusion detection — produced multiple peer-reviewed publications.
- Cross-feed correlation using data-mining and ML
- Published work on intuitionistic-fuzzy kernel clustering for forensics
- Co-authored "Towards Automated Threat Intelligence Fusion"
- Research
- Security
- Django
- MongoDB
04 / Experience
Thirteen years across applied ML and software.
- Sep 2021 — PresentCupertino, CA
Machine Learning & Engineering · Apple
Lead the technical roadmap for a centralized anomaly-detection & data-quality ML platform; founded an InnerSource PMC; led PII detection across 100% of warehouse columns; designed an enterprise anti-fraud system; spoke at Databricks Data+AI Summit 2025.
- Aug 2019 — Sep 2021San Francisco Bay Area
ML Lead, Search Ranking · Walmart Global Tech
Founding ML engineer & research lead for Walmart.com's reranking team. Invented the Rerank Micro Service (RMS), Walmart's first microservice for reranking. Shipped Learning-to-Rank and embedding-based semantic matching tied to nine-figure revenue impact.
- May 2017 — Aug 2019San Francisco Bay Area
Senior Machine Learning Engineer · Yahoo
Built a stock recommendation system for 1M+ daily users (Spark MLlib), an analyst credibility scoring system, and a CNN-based event classifier for Yahoo Mail. Productionized a hierarchical classifier handling 100M+ emails/day, and led Resource-as-a-Service for real-time model fetching.
- May 2015 — May 2017Tempe, AZ
Machine Learning Engineer · Arizona State University
Designed Threat Intelligence Analytics — a platform correlating security data across heterogeneous feeds using data-mining and ML. Yielded multiple peer-reviewed publications on intuitionistic-fuzzy clustering for intrusion detection.
- 2016
Software Engineer Intern · Goldman Sachs
Summer internship building internal engineering tooling.
- 2014 — 2015Bangalore, India
Software Engineer · Symantec
Big-data security analyst — captured incident vectors from IDS / endpoint products, stored them in HBase, and classified malicious activity with data-mining and ML on Hadoop, Hive, and IBM InfoSphere Streams.
- 2013 — 2014Bangalore, India
Software Engineer · EMC
Built tooling and diagnostics for EMC Documentum / xCP. Owned the Documentum Mobile 1.2.4 iOS release end-to-end. Bronze Award, Value Creation, Q1 2014.
05 / Research & Speaking
Publications, talks, and recognition.
Speaking
Databricks Data+AI Summit 2025
Selected from a record number of submissions
Recognition
Senior Member, IEEE
Plus EMC Bronze Award · Hack Arizona winner
Mentorship
Career mentor
Onboarded 7+ engineers; multiple to feature ownership
Selected publications
View on Google Scholar- 01
Reasoning-based Anomaly Detection (preprint)
Anomaly detection · LLM reasoning · 2025
- 02
Towards Automated Threat Intelligence Fusion
Threat intelligence · multi-source correlation
- 03
A Kernel based Atanassov's Intuitionistic Fuzzy Clustering for Network Forensics and Intrusion Detection
Network forensics · intrusion detection
- 04
Evaluation of Kernel Based Atanassov's Intuitionistic Fuzzy Clustering for Network Forensics and Intrusion Detection
Network forensics · empirical evaluation
- 05
An Intuitionistic Fuzzy Kernel Clustering for Medical Image Segmentation
Medical imaging · segmentation
- 06
A Novel Intuitionistic Fuzzy Neural Network approach for solving Higher Order Fuzzy Polynomials
Fuzzy systems · neural networks
06 / Contact
Let's build something useful.
I'm always happy to hear from folks working on hard ML or data-platform problems — whether it's a role, a collaboration, or just a conversation about anomaly detection, search ranking, or applied AI. I also mentor early-career engineers; if that's you, please don't hesitate to reach out.
anupam.p.iitd@gmail.comUsually reply within a day or two