Open to opportunities

Harika Yenuga

Senior AI/ML Engineer shipping
RAG that actually retrieves

Currently leading GenAI at Macy's. Previously caught fraud at Bank of America and built retail ML at Apple. I ship AI that works Monday morning, not just in notebooks.

United States
8+
Years Deep in ML
$2B+
Protected Daily
99.9%
Uptime Shipped
About Me

The Story Behind The Code

I spent 8 years learning that the hard part of AI isn't the model—it's everything else. Data quality. Latency. Edge cases. Stakeholder buy-in. Prod incidents at 3am. Built retail forecasting at Apple, fraud detection at Bank of America, and now GenAI infrastructure at Macy's. Each role taught me that shipping beats perfecting.

End-to-End Ownership

I own the full stack: data pipelines, model training, deployment, monitoring. No handoffs, no "that's not my job." When something breaks at 2am, I fix it.

Business First

I start with the problem, not the tech. Stakeholders don't care about your architecture—they care if it works. I translate ML capabilities into business outcomes.

Production or Nothing

Jupyter notebooks don't count. I measure success by what's running in prod, handling real traffic, making real money. MLOps isn't overhead—it's the job.

Where I Learned

The foundation of my journey

MS Business Analytics

Northwood University • GPA: 3.87/4.00 • 2024

B.Tech Electronics & Communications

JNTU • 2015-2019

Moments That Mattered

Milestones along the way

Research Publication

Predictive Hiring with ML • IBAC May 2025

Spot Awards (x2)

Apple Inc. • Exemplary Performance

Experience

Four Chapters,
One Journey

Every role taught me something different. Together, they shaped how I think about building AI that matters.

Macy's Inc.

Current

Lead AI Engineer

GenAI Platform & Applied ML Systems

Mar 2025 - PresentUnited States
Writing the next chapter: bringing GenAI to 500+ retail stores

This is where everything comes together. After years of building ML systems, I finally get to lead a GenAI transformation from scratch. The challenge is not just technical but cultural: how do you help 500+ stores trust and adopt AI? Every day brings new questions, and I love that.

Started with a blank canvas. Built the entire LLM platform from zero, designing RAG architecture that now answers questions across millions of product documents in under 2 seconds

Sub-2s Latency

Created autonomous AI agents that actually think. They decompose problems, pick the right tools, write SQL, and correct their own mistakes. Watching analysts save 60% of their time never gets old

60% Time Saved

Built trust through safety. Our guardrails ensure 99.9% of responses are safe, because in enterprise AI, one bad answer can undo months of adoption work

99.9% Safe

Made deployment boring (in the best way). CI/CD, versioning, monitoring all working seamlessly so the team can focus on building, not firefighting

Production MLOps
LangChainLangGraphAmazon BedrockFAISSSparkDatabricksFastAPIPython

Bank of America

Senior AI Engineer

ML Engineering & Financial Risk Systems

Feb 2022 - Jan 2024United States
The chapter where stakes got real: protecting $2B+ in daily transactions

Banking taught me what it means when your code really matters. When your model flags a transaction as fraud, you are either saving someone their life savings or blocking a legitimate purchase. That pressure shaped how I think about every system I build now.

The fraud problem was personal. Built models that cut false alerts by 35% while catching more actual fraud. Behind those numbers are real people who kept their money or did not get wrongly blocked

35% Fewer False Positives

Learned to think in petabytes. Built pipelines processing more data in a day than most companies see in a year. The scale was humbling and addictive

PB-Scale Engineering

Got obsessed with deployment speed. Reduced model releases from weeks to hours because in fraud detection, every day of delay means more money lost

Hours Not Weeks

Led the GenAI charge before it was cool. Built document AI systems that 200+ analysts now rely on daily. Seeing skeptics become believers is still the best feeling

200+ Daily Users
PythonXGBoostPyTorchPySparkAzure OpenAISnowflakeBigQueryDockerMLflow

Apple Inc.

AI Engineer

Retail Analytics & ML Systems (via Infosys)

Sep 2019 - Feb 2022United States
Where I learned what excellence looks like: Apple Retail analytics

Apple changed how I think about quality. Their bar for everything is impossibly high, and it is contagious. I came in to build dashboards and left knowing how to lead teams, ship ML systems, and demand excellence from myself.

My dashboards reached executives. Built pipelines processing 50M+ daily transactions so leadership could see every store on the planet in real-time. No pressure, right?

50M+ Daily Records

Convinced the team to go from reports to predictions. Those forecasting models cut stockouts by 25%. Somewhere, a customer got the iPhone they wanted because of math we wrote

25% Better Forecasts

Discovered I love teaching. Mentored 19 engineers and watched them grow. Turns out the best way to learn is to help others learn

19 Engineers Mentored

Two Spot Awards taught me something important: ship quality work ahead of schedule, and trust follows. Those awards opened doors to bigger challenges

2x Recognition Awards
PythonSQLAzure MLSageMakerProphetFastAPIDockerKubernetesTableau

RSI Softech

Data Engineer

National Remote Sensing Centre (Govt. of India)

Jun 2017 - Aug 2019Hyderabad, India
Chapter one: where satellite images sparked a data obsession

Every journey starts somewhere. Mine started with satellite imagery from space and a lot of messy data. I did not know it then, but processing images of cities from orbit would teach me to love the puzzle of making sense of massive, complex data.

Built my first terabyte-scale system. Designing geodatabases for Smart Cities felt like science fiction. Watching urban planners use data I processed to improve real cities? Pure magic

TB-Scale Design

Discovered automation is my superpower. Cut manual processing by 70% and suddenly had time to learn, experiment, and dream bigger

70% Faster Processing

Learned that garbage in means garbage out. Taking data quality from 60% to 95% taught me to obsess over the fundamentals. That lesson still guides everything I build

95% Data Quality
ArcGISQGISPythonPostgreSQLMySQLSQL ServerETL
8+
Years of Growing
4
Chapters Written
50+
Stories Shipped
2
Awards Earned
Tools of the Trade

Technologies I Think In

Every tool here has a story. Some I learned through late-night debugging. Others through production incidents. All of them shaped how I solve problems today.

Where I Live Now: GenAI

LangChainLangGraphRAG PipelinesPrompt EngineeringFine-tuning (LoRA, QLoRA)Claude (Bedrock)OpenAI APIsGeminiFAISSPineconeSemantic Kernel

The Foundation: ML

PyTorchTensorFlowXGBoostLightGBMScikit-learnDeep LearningNLP PipelinesTime-seriesSHAPLIME

Moving Data at Scale

Apache SparkPySparkDatabricksETL/ELTDelta LakeKafkaFeature StoresData ModelingSnowflakeBigQuery

Making It Run Anywhere

AWS (SageMaker, Bedrock, EMR)Azure (Databricks, ML)GCPDockerKubernetesCI/CDMLflowGitHub Actions

Languages I Dream In

PythonSQLPySparkRFastAPIREST APIsTypeScriptGit

Making AI Understandable

Model ExplainabilitySHAP VisualizationsAttention MapsEmbedding ProjectionsStreamlit DashboardsGradio Interfaces

Where Data Lives

PostgreSQLMySQLMongoDBSQL ServerOracleRedshiftHBase

The Future: AI Agents

Multi-Agent WorkflowsTool OrchestrationGuardrailsCrewAIEvaluation FrameworksA2A Protocol
Side Projects & Research

Curiosity ProjectsThat Became Real

These started as "what if" questions on quiet evenings. Some became research papers. Some became tools people use daily. All of them taught me something I could not learn at work.

Featured Project

MyFoodTracker

A Personal Problem That Became a Platform

It started with frustration. Existing nutrition apps felt slow and inaccurate. So one weekend, I decided to build something better. Six weeks later, I had a full AI platform: snap a photo, get instant nutrition data. The journey from "I can do this better" to "people are actually using this" was the most satisfying thing I have built.

TypeScriptReactNode.jsTensorFlowD3.jsDocker

What I Learned Building This

Trained my own vision model on 100K+ food images. Watching it correctly identify a complex salad for the first time felt like magic
Built a recommendation engine that learns what you like. Users tell me it feels like the app "gets" them
Created dashboards that make data feel personal. Seeing your progress visualized keeps people coming back
Made it production-ready because half-finished projects teach you nothing. CI/CD, Docker, auto-scaling, all of it
Blog & Research

Thoughts & Publications

Sharing knowledge and insights on AI, ML, and data science

Research Publication

Revolutionizing Recruitment: Enhanced Machine Learning Models for Bias Mitigation and Efficiency

International Business Analytics Conference (IBAC) 2025

Submitted to Empirical Economics Letters (EEL)

Engineered end-to-end predictive hiring models achieving 97% accuracy while reducing bias in candidate selection.

View on GitHub
The Next Chapter

Let's Write the Next Story

I'm looking for my next chapter: a team tackling hard problems with AI, where I can learn, contribute, and help ship things that matter. If that sounds like what you're building, I'd love to hear your story.

Every Great Thing Starts With a Conversation

Whether you are hiring, exploring a collaboration, or just want to talk about the future of AI over coffee, I answer every message. The best opportunities often come from unexpected conversations.

Ready for my next chapter