Hello and welcome to my page! My name is Marika Lam and I'm a Senior Software Engineer at Walmart Global Tech, based in the Silicon Valley. I lead engineering for Walmart's Promotions and Coupons platform and am building PromoGenie, an agentic AI platform that uses Claude and RAG-based retrieval to streamline promotion and coupon workflows. Before Walmart, I graduated from Boston University and worked at Rakuten in San Mateo.
In my free time, I enjoy yoga, playing tennis, cooking and anything that requires being outdoors. I also love writing tutorials and articles on my personal Medium page. You can see my apps and projects under the "Apps & Projects" section.
• Lead engineering efforts for Walmart's Promotions and Coupons platform, building internal products that enable business teams to create, manage, and execute promotions across 11 international and U.S. markets. • Serve as a tech lead for full-stack initiatives, driving architecture, design reviews, implementation, and modernization of promotion-management experiences using React, TypeScript/JavaScript, Java, Python, Node.js, REST APIs, and distributed backend services. • Develop PromoGenie, an agentic AI platform that streamlines promotion and coupon workflows using LLMs, Claude, Python, TypeScript, ChromaDB, vector search, and RAG-based retrieval. • Partner with product managers, backend engineers, platform teams, and business stakeholders to translate complex promotion rules and operational requirements into scalable product and technical solutions. • Provide architectural guidance and technical leadership across the team, mentoring engineers, reviewing designs and code, and driving engineering standards for maintainability, performance, testing, and developer experience. • Built automated end-to-end and API testing frameworks using Playwright and Nightwatch.js, expanding coverage across the Promotions UI and 30+ REST APIs. • Worked on backend and data engineering for the Promotions platform, building services with Java, Spring MVC, REST APIs, Oracle, and Cassandra.
Worked as a backend Java and database engineer on the Promotions platform before transitioning into a full-stack role, building RESTful services with Java, Spring MVC, Oracle, and Cassandra.
Worked on back-end infrastructure and services of the Ecosystem Department and Membership team by creating scalable APIs to be used by clients and its subsidiaries: RESTful web services (Java, Spring, Cassandra, RabbitMQ); batch processing and migration from Oracle to Cassandra (Scala, Spark, Java, Python); integration, unit and performance tests (Cucumber, JUnit, Gatling); development process organization (Confluence, Jira, Jenkins, Git); and code reviews and quality control (Git, Atlassian Stash, Jira). Worked at the Tokyo Headquarters office in 2015.
Lubax is a smartphone-based clinical reference to identify skin lesions. The user takes a photo of a skin lesion and the app will search a proprietary image database of diagnosed lesions to identify a diagnosis. CBIR (content based image recognition) is used to discriminate skin-lesion types. I data mined 15,000+ image files of skin diseases for the iOS platform in Java and Python, and designed and developed the iOS app using Objective-C, C++, and the XCode IDE.
Retrieved endowment-related data from DARWIN (Development and Alumni Relations Web-Based Information Network) database. Analyzed and validated reports before sending to sales and marketing, revenues, and other departments.
At BU, I was involved in many groups and activities. I worked as an Admissions Ambassador, leading campus tours for incoming students and parents, as well as hosted Open Houses.
I was also a member of Alpha Phi Omega, a community service fraternity, and completed 80+ service hours per semester. I spent my weekends volunteering at the Red Cross, Dana-Farber Cancer Institute, the Boston Food Bank, and various other places. I also enjoyed programming (not related to my school projects), so I joined Global App Initiative, where I created various iOS apps for non-profit organizations.
HomeDiary is a private diary for your home. Sign in, add your properties, and log a timeline of everything that happens to them: purchase, damage, repairs, inspections, and more, each with photos and documents attached. Paperwork like purchase documents, insurance policies, and warranties gets its own place for each property, and every account only ever sees its own homes. The app has a typed REST API with JWT cookie auth and bcrypt password hashing, and uploads go straight to S3-compatible object storage. It runs as a single Docker service on Render.
Take a photo of your notes or paste them in, and KachunQuizer.ai turns them into an Anki-style flashcard deck. Tesseract.js reads the text out of the photo in the browser. Claude then writes real comprehension questions from it, through a Cloudflare Worker that keeps the API key off the client. If that request fails, the app falls back to an offline generator that uses NLP to blank out the key word in each sentence, so a study session never hits a dead end. Each deck uses spaced repetition: facts you miss come back sooner, and facts you know show up less often.
MathPop is a sibling to PitchPop that uses arithmetic instead of chords. Kids pick multiplication, addition, subtraction, or a random mix, then solve five-problem rounds with big colored cards, chimes, and confetti. They can tap an answer or write it by hand: a small neural network I trained on MNIST, with augmentation to handle messy kid handwriting, reads the digits right in the browser. The wrong answer choices come from common slip-ups, like being off by one factor or adding instead of multiplying, so getting it right means knowing the answer. Problems are read aloud with Piper TTS, and progress is saved on the device and broken down by operation.
PitchPop started as a game we play at home: I play a chord on the piano, and my kids grab the matching color flag and shout it out. The app turns that into a quiz. Tap the rainbow to hear one of nine chords, pick its color from four choices, and get confetti and spoken feedback when you're right. The piano sound is synthesized from harmonics with the Web Audio API, so the app ships no audio samples. The voice is Piper, a neural text-to-speech model that runs in the browser through ONNX Runtime. There's also a free-play mode, plus a progress page for each child that tracks accuracy and which colors have come up. It installs to a phone's home screen as a PWA.
Card payments cost merchants about 2.9% per transaction, and paying straight from a bank account online usually fails because there's no real-time check that the payment is good. PowerPay authenticates the payer and pays with a digital check drawn on their bank account, using the Feedzai API to score each payment for fraud before it goes through. I built the web app in Flask and deployed it on Heroku.
4Sight helps small businesses manage working capital by forecasting cash flow by day, week, or month. It starts from a worst-case projection that assumes every customer pays on the last possible due date, then updates in real time by checking the business's bank account through the US Bank API, so invoices that are paid early move the forecast right away.
At Changi Airport, 60% of passengers arrive less than two hours before their flight, and even early travelers worry about missing it. Changi Along The Way predicts how long it will take to get to your gate and suggests a personalized list of places to stop along the way, so you can relax without losing track of time.
Roomz is an iOS app for checking, booking, and canceling meeting rooms at Rakuten's Tokyo office. It shows room availability in real time through the Microsoft Office 365 Outlook API, so people could grab a free room on the spot and the company didn't need to pay for a third-party booking tool.
Photos on review sites only show part of a restaurant, and you can't tell how big a dish really is. Tolen lets you experience a restaurant before you go: browse a clickable menu that shows each dish as a hologram, and look around the dining room in a 360-degree video. Restaurants can show off their best-rated dishes, and diners know what to expect.
BUSale is a free marketplace for college students. Anyone can sell a used item or a service by posting it on Instagram with #busale in the caption, and buyers browse and buy those listings on the BUSale website. It pulls posts through the Instagram API into an MVC web app written in PHP with a MySQL database.
FIF was the first iOS app built for the Boston University School of Medicine. It's a medical education app that helps students practice finding the resources behind evidence-based medicine. I designed it with Balsamiq wireframes, built it in Objective-C, beta tested it through TestFlight, and shipped it to the App Store.
I'm open to new projects, freelance work and collaborations. Have an idea or need a hand building something? Send me a message below or email me directly at marikaleelam@gmail.com.