Rithvik Praveen Kumar

• Selected for a faculty-advised, credit-bearing research team of 30+ students building the full autonomy stack, perception, localization, planning, controls, and simulation for Purdue's driverless racing platforms.
• Working within the software/AI track to apply machine learning and computer vision methods to on-vehicle autonomy problems, developing in Python and C++ on a ROS 2 codebase with LiDAR, camera, and GNSS/IMU sensor data.
• Scoping a semester-long research project with faculty and graduate-student mentors, to be validated in simulation against logged track data before deployment on the team vehicle.

PythonC++ROS 2Machine LearningComputer VisionSensor Fusion (LiDAR, Camera, GNSS/IMU)

• Developing FieldWatch 2.0 console with DELMAR Software, rebuilding the DriftWatch specialty crop registry admin experience into a full operations console used by producers, beekeepers, and pesticide applicators across multiple states.
• Rebuilding admin workflows, report generation, bulk export, and user administration, on a Python & PostgreSQL backend with a Vue front end
• Designing API access control and service boundaries as the platform moves toward a micro-services architecture
• Partnering with DELMAR product owners to translate operational requirements into system design, contributing architecture proposals on services, API contracts, and schema changes
• Improving query performance and data accuracy across the registry's mapping and reporting paths
• Delivering on a milestone cadence (Agile Software Development Methodology) with demos and prototypes reviewed directly by the DELMAR engineering team.

PythonPostgreSQLJavaScriptVueREST API designmicro-service based architectureAgile Project ManagementDatabase DesignApplicationProgrammingApplication Programming Interfaces (API)Software ArchitectureFull-Stack DevelopmentGitTechnical CommunicationRequirement AnalysisContinuous Integration and Continuous Delivery (CI/CD)Query OptimizationSQL

• Researching TinyML systems for edge inference under Prof. Qiuyue "Shirley" Xue.
• Optimizing and deploying small ML models onto low-power embedded hardware via Edge Impulse.
• Investigating novel low-power designs, including piezoelectric microphones that act as both sensors and energy sources for on-device models.
• Weighing model accuracy against hardware and energy constraints through implementation, experimentation, and literature review.

TinyMLEdge ImpulseEmbedded MLPythonC/C++Edge Inference

• Built a full-stack, AI-powered RAG learning platform for a non-profit serving 500+ underserved learners studying personal finance.
• Wired the RAG pipeline, LLM workflow, and vector retrieval that grounds chatbot answers in actual lesson content.
• Added AI generation of quizzes and assignments, and shipped the core lesson-delivery UI and backend.
• Translated non-technical product requirements into technical specs, working with designers and stakeholders end-to-end.

ReactNode.jsExpressMongoDBOllamaRAGVercelCloudflare

• Promoted role layered on top of Code Sensei duties.
• Analyzed company revenue, costs, and margins to surface historical trends and key financial insights.
• Delivered quarterly reports to colleagues and associates.
• Democratized investment knowledge across the organization by fielding policy questions and teaching peers how to read company analysis.

Financial AnalysisReportingInvestor Education

• Taught programming through game development to 40 students aged 5–12, covering JavaScript, C#, MakeCode, and 3D modeling/printing.
• Built personalized skill-development paths with parents and ran demos for prospective families.
• Organized hackathons and kept the learning environment positive, safe, and a little chaotic in the best way.

TeachingJavaScriptC#MakeCode3D ModelingMentorship

A robotic arm driven by thought alone

• Drives a robotic arm from thought alone — a proof-of-concept brain-computer interface.
• Pairs a NeuroSky EEG headset with an Arduino microprocessor.
• Built as a foundation for future BCI work translating intent into sign-language gestures for differently-abled users.

NeuroSkyArduinoC++PythonEEG

Demo ↗

Finds every frame a person appears in a video

• Full-stack web app that pinpoints every appearance of a target person in a video.
• Ingests reference images and footage, then runs a facial-recognition pipeline.
• Built on the MERN stack with AWS Rekognition.

ReactExpressMongoDBAWS RekognitionMERN

GitHub ↗

Classifies left, right and neutral thoughts from EEG

• Classifies left, right, and neutral thoughts from 16-channel EEG data captured with an OpenBCI headset.
• Pairs a stimulus-display program with TensorFlow models.
• Ongoing work focuses on dataset collection and classification accuracy.

PythonOpenBCIBrainFlowTensorFlowKerasEEG

GitHub ↗

Modeled how public spending tracks happiness

• Trained a Lasso model to correlate national spending with reported citizen happiness.
• Analyzed spending across military, education, and health.
• Built by a 4-person team during Columbia University's SHAPE program in NYC.

PythonPandasscikit-learnLassoSHAPE

Demo ↗GitHub ↗

Turns any text into a downloadable QR code

• Web app that turns user text into a downloadable QR code.
• Python and Segno backend with a hand-rolled HTML/CSS/JS frontend.
• Built to practice the full deploy loop, hosted on an AWS EC2 instance.

PythonSegnoHTMLCSSJavaScriptAWS EC2

GitHub ↗

One-click listings to Marketplace and eBay

• Built and rolled out an inventory-listing tool for Habitat for Humanity.
• Posts items to Facebook Marketplace and eBay from a single interface.
• Authored training materials and onboarded staff to the new workflow.

Google Apps ScriptJavaScriptAutomation

GitHub ↗

B.S. in Computer Science + Mathematics · Software Engineering · AI/ML