Daniel Hariyanto

Microsoft

Redmond, WA   |   MAY 2022 - AUG 2022

Product Manager Intern

• Investigated new business areas to expand Microsoft's content delivery optimization technology, which is responsible for reliably and performantly distributing >10 exabytes of Microsoft content per month to over 1.4 billion Windows devices globally
• Designed a go-to-market plan that evaluates business opportunities and risks with 3rd party content support for Microsoft's peer-to-peer and caching services, addressing highly expressed customer pain points (88% customer approval via surveys) and providing value-add to Microsoft's Intune and Azure CDN services simultaneously
• Conducted usage and revenue modeling based on large internal datasets across 3 teams to evaluate growth opportunities in the next 5 years
• Created feature spec and UX wireframes using Figma, coordinated feature implementation with the development team, and presented project to CVP

Bessemer Venture Partners

San Francisco, CA   |   FEB 2022 - AUG 2022

Bessemer Fellow & Software Engineer Intern

• 1 of 20 Fellows selected out of >2,000 applicants across the US to be part of the Bessemer VC network - learn more
• Interned as a software engineer at Bevi, a Bessemer portfolio company, building Alexa smart speaker functionality for Bevi's IoT flagship product
• Built a proof-of-concept interface based in Java Dropwizard that enables a custom Alexa skill to connect to Bevi's IoT product and recreate all functionality via voice commands; created detailed Confluence documentation on the interface's endpoints and on Alexa skill creation and distribution

• Learn more

PayPal

San Jose, CA   |   MAY 2021 - AUG 2021

Software Engineer Intern

• Implemented features in web application of PayPal's Business Debit Mastercard, the #1 largest US Mastercard small business debit portfolio; built multiple secondary cards functionality and recurrent payments dashboard with React.js & Node.js (Kraken framework) used by >1.8 million cardholders
• Designed and led isolated component testing in Java Spring application of PayPal's Shipping & Tracking, which reduced dependency on shared staging environments and enabled developers to have complete control over their test suite; increased development productivity by >400%
• Developed NLP clustering model for PayPal's Innovation Lab that classifies ideas based on their content; after receiving positive feedback from leadership, implemented model in their Idea Portal now used to classify >4,000 yearly instances of employee ideas to inform better decision-making for key stakeholders

Hong Research Group

(a machine learning research group within the Department of Computer Science at Brandeis University with a focus on ML applications)

Waltham, MA   |   AUG 2022 - MAY 2023

Undergraduate Researcher

• Spearheaded independent machine learning research with Dr. Pengyu Hong, investigating novel dataset pruning strategies that enabled accelerated training convergence while maintaining remarkable accuracy; novelty: uniquely targeted samples with similar difficulty levels for removal
• Successfully achieved a notable >30% reduction in training time with a mere 0.3% accuracy compromise, validated through rigorous trials with ResNet-18
• Impact: more time-efficient training processes, reduction in computational resource requirements, and enhanced accessibility to previously large datasets

Brandeis Online Learning Lab

(a cutting-edge classroom analytics tool to measure student engagement in online classrooms, spearheaded by Professors Pito Salas and Ben Gomes-Casseres)

Waltham, MA   |   MAY 2021 - MAY 2023

Software Engineer

• Developed an end-to-end data pipeline that transforms Zoom class recording data into useful data abstractions and visualizations; this is a university-funded research project exploring how pedagogy affects student engagement in online courses and, in particular, how it can enhance equity in the classroom
• Built PostgreSQL database within on-prem Apache server from scratch that models complex interactions among courses, instructors, Zoom meetings, participants, and transcripts
• Designed efficient algorithms capable of matching >80,000 Zoom meetings to classes and >60,000 encrypted Zoom participants to registered students
• • Leveraged advanced SQL join queries and analytical algorithms to transmute raw Zoom data into insightful statistics on student engagement

• Learn more

TAMID Group

(a student-run organization that facilitates hands-on experiential learning with Israeli companies for college students)

Waltham, MA   |   AUG 2020 - MAY 2023

Director of Tech Consulting & President

• Lead 11 student engineers across two teams to complete technical consulting projects for three Israeli tech startups: Mappo, SOLO, TapTapes, & Vibo
• Spearhead client-project matching, project scoping, client expectation negotiations, user story creation, AGILE practices, and team communications
• Ensured quality of >300 total commits through conducting code reviews that have been pushed to production; conducted weekly presentations to client
• Consistently achieved perfect 5-star evaluations from previous clients, by often outperforming clients' expectations and delivering high quality of work
• Created over 4 weeks of education curriculum on Git, HTML/CSS, JavaScript, React.js, React Native, Node.js, GraphQL, MongoDB, & Docker

Mappo

(an in-car app for location-based cultural content working with leading automotive OEMs)

Tel Aviv, Israel   |   FEB 2022 - MAY 2022

Software Engineer

• Led a 5-person team to build book collection APIs using BeautifulSoup, Flask, and Heroku for Mappo's literature-based content
• Designed clever CAPTCHA-bypassing web scrapers for Amazon, Goodreads, and Project Gutendex that return cleaned data for >428 million books
• Built Mappo's book data insertion pipeline, which will be responsible for providing >70% of Mappo's cultural content in their app

SOLO

(an emotional wellbeing app that uses facial recognition to measure mood cues)

Tel Aviv, Israel   |   SEP 2021 - DEC 2021

Software Engineer

• Led a 3-person team to build an iFrame widget implementation of SOLO's flagship product based in React.js frontend and Node.js backend
• Enabled SOLO's product to exist as an isolated app widget on external websites, which is currently used by one of SOLO's clients with >200k users

TapTapes

(a podcast-sharing social network mobile app)

Tel Aviv, Israel   |   FEB 2021 - MAY 2021

Software Engineer

• Led a team of 4 student engineers to build monetization strategy of TapTapes, a podcast-sharing social network mobile application
• Designed digital tokenization mechanism in React Native frontend and Node.js backend with MongoDB database and S3 storage, to become TapTapes's first revenue stream

Vibo

(a music management tool for DJs and event planners)

Tel Aviv, Israel   |   AUG 2020 - DEC 2020

Software Engineer

• Led a team of 2 student engineers to create a song trends analysis platform on Vibo, a music management tool for DJs and event planners, based in React.js frontend and Node.js & GraphQL backend that queries MongoDB database; increased screen-time and user engagement by >40%
• Headed software testing: internal testing using generated fake data via Faker.js & GraphQL integration tests via Jest

AgroFides

(an analytical tool to assess creditworthiness for African farmers)

Newburyport, MA   |   NOV 2020 - JAN 2021

Software Developer & Risk Analyst Intern

• Investigated psychometric metrics to predict risk behavior of Ghanaian farmers to assess their creditworthiness for microlending
• Built psychometric test web application with Vue.js & Nuxt.js frontend and Flask backend to assess risk indicators in >1,200 Ghanaian farmers
• Designed unsupervised ML model that identifies farmers with low-risk, creditworthy attributes from psychometric test results

W. B. Rogers Lab

(an experimental research group within the Department of Physics at Brandeis University working in the areas of soft matter and biological physics)

Waltham, MA   |   JUN 2020 - JUN 2021

Undergraduate Research Fellow

• 1 of 7 to acquire two highly competitive undergraduate research fellowships, funded by the NSF and by Brandeis University, for two consecutive years
• Organized independent nanotechnology research with Dr. Ben Rogers and Dr. Melissa Rinaldin, analyzing the liquid-liquid phase separation (LLPS) behavior of DNA nanostars
• Investigated unexpected LLPS behavior of DNA nanostars at high temperatures using a variety of experimental techniques including confocal, brightfield, & DIC microscopy (using Leica Application Suite X), FRAP, and NUPACK
• Conducted extensive data analysis using a variety of NumPy, Pandas, Matplotlib, Scipy, and image-analysis Python libraries to investigate the biophysical properties of DNA nanostars (see here)

Mobile Memory

Early-Onset Alzheimer's Detection using Speech

• Lead and coordinate development of iOS/Android mobile application that analyzes vocal biomarkers to screen for early signs of dementia & Alzheimer's
• Constructed cloud data pipeline and state-of-the-art machine learning model using actual audio datasets of people with dementia from DementiaBank
• Currently an early-stage MassChallenge venture that is undergoing clinical validation at Swedish BioFINDER study in collaboration with the Carney Institute of Brain Sciences at Brown University
• Won multiple awards & grants: Brandeis SPARKTank Competition (1st place), AI@MIT Generator Competition (2nd place), Harvard Capital Partners Fund Winner, Dove Chocolate InstaGrants Winner (1 of 3 to win), Harvard iLab Spark Grant, Rothberg Catalyzer Prize Finalist, & Harvard CBE Venture Finalist

Technologies: React Native (frontend), Firebase (backend & cloud storage), Firestore (database), BERT + wav2vec model (ML)

• Learn more

Novozymes Enzyme Stability Prediction

Kaggle Competition

• Achieved top 7% placement (Leaderboard Team Name: The Brogrammers); predicted thermostability of enzyme variants based on protein sequence
• Designed an ensemble model with BLOSUM, pLDDT, DeepDDG, DeMask, RMSD, SASA, Rosetta, and ThermoNet models that achieved a Spearman's correlation coefficient of 0.606 compared to ground truth

Technologies: BLOSUM, pLDDT, DeepDDG, DeMask, RMSD, SASA, Rosetta, ThermoNet, PyTorch

• Learn more

Shopee Product Matching based on Image and Text Description

Kaggle Competition

• Achieved top 8% placement (Leaderboard Team Name: Shopee Masters); determined if two products are the same based on image and text descriptions from Shopee's e-commerce dataset
• Designed an ensemble model of EfficientNets B2-B6 with ArcFace loss for image embeddings, TF-IDF vectorizer for text embeddings, and similarity score function from the embeddings using k-nearest neighbors; achieved F1-score of 0.734

Technologies: EfficientNets (ensemble) for image embeddings, TF-IDF vectorizer for text embeddings, k-nearest neighbors, PyTorch

• Learn more

HeatPumpBuddy

Resource Hub for Prospective Heat Pump Owners

• Built interactive web application for local volunteer group HeatSmart Alliance to incentivize homeowners to replace current high carbon-emitting systems with energy-efficient heat pumps; currently used by HeatSmart Alliance to reach >14,000 homeowners across Massachusetts
• Designed a heat pump calculator to help homeowners calculate annual heat pump costs, and a networking functionality that connects homeowners to heat pump coaches based on house, location, and lifestyle similarity, driven by a nearest neighbor algorithm
• Won 1st place and the Highest Organizational Impact prize at Deis Hacks 2021, a social good hackathon for local organizations

Technologies: React.js + Next.js + Redux (frontend), Flask (backend), nearest neighbor algorithm + scikit-learn (ML)

• Learn more

iPresent

AI-powered Public Speaking Evaluator

• Created web application that provides both a real-time presentation simulator and evaluator to improve users' public speaking skills
• Engineered a set of algorithms based on multi-label classification, CNN, and GC speech & language APIs that utilizes transcript and audio as training modalities to compute useful presentation metrics such as rating classification, brevity, cadence, diction, etc.

Technologies: Vanilla JS (frontend), Flask (backend), Google Cloud Storage (database), Google Cloud speech & language APIs + CNN + TensorFlow (ML)

• Learn more

MediLoc

Optimal Location Finder for Nutrition Centers in Rural Indonesia

• Built web application that identifies optimal locations for nutrition centers to aid the Indonesian government combat stunted growth in rural villages in Indonesia
• Developed backend in Flask and optimal location algorithm based on k-means clustering and population density

Technologies: React.js (frontend), Flask (backend), SQLAlchemy (database), k-means clustering + scikit-learn (ML)