01. About
A bit about me
I'm a Computer Science student at the University of Maryland, College Park, with a minor in Business Analytics. I build production AI pipelines and full-stack systems, from multi-agent tax filing systems to document intelligence tools processing real government workflows.
I care about shipping things that actually work under real constraints, not demos. My work spans agentic orchestration, RAG pipelines, and traditional full-stack development, and I've had the chance to apply that across a VA claims processing pipeline, an enterprise GenAI platform at EY, and a first-place hackathon win.
Outside of engineering, I'm part of the AI/ML Club and Competitive Programming Club at UMD, and I spent a semester studying at Universidad Carlos III de Madrid.
Multi-Agent Systems
Orchestration
RAG Pipelines
LLM Engineering
Full-Stack
FastAPI · React
Cloud
AWS · Snowflake
02. Experience
Where I've worked
- ▸Built a full-stack GenAI platform (FastAPI, React, LangChain) that extracts and enriches enterprise KPIs from unstructured client documentation at 89% extraction accuracy, cutting per-client turnaround from roughly a month of manual analysis to a matter of days.
- ▸Shipped to production and adopted by consultants across the engagement team, processing 100+ clients in its first two months against a manual baseline of roughly one client per month.
- ▸Designed a 3-stage validation engine and chunked Map-Reduce pipelines so documents exceeding a single model context window process in parallel and recombine without losing cross-section references.
- ▸Automated the downstream design step end to end: requirement documents in, structured Snowflake DDLs and SQL transformation logic out, removing a manual schema-authoring stage from delivery.
Interactive
Try out a couple of my projects
Both of these run entirely in your browser: no sign-up, no upload, no server. Open either one for the full write-up and a demo you can put your own documents through.
03. Projects
Other things I've built
Keepsake Diary
Co-Founder
A 3D-printed interactive diary for young Harry Potter fans, powered by a locally fine-tuned open-source LLM with custom safety guardrails. Owned product and technology, driving 30M+ views and 1.2M+ likes on launch content, building a waitlist of 1,000+ customers and fulfilling the first 20 orders.
Muá! Specialty Coffee & Brunch
Freelance / Contract Web Development
Designed and built a fully responsive, bilingual (ES/EN) marketing website for an independent café in Madrid, Spain, with a custom i18n system, reservation flow via Instagram DM deep-linking, and a token-based design system.
G.E.O.P.A.L
🏆 1st place, HopHacks 2023 (30+ teams)
Full-stack environmental data platform integrating Google Earth Engine APIs to surface satellite-derived insights for NGO decision-making, with OpenCV facial-recognition auth and tamper-proof identity verification via the Verbwire API, minting face-data NFTs on-chain.
Regulatory Document Intelligence Tool
End-to-end RAG pipeline ingesting regulatory PDF reports and extracting structured environmental data, with natural language querying over ingested documents via vector embeddings.
Panorama Stitching
Computer vision pipeline implementing ANMS keypoint selection, RANSAC-based homography estimation, and feature matching to stitch 4+ image sequences into seamless panoramas with sub-pixel alignment accuracy, tuned across 20+ indoor/outdoor test sets.
Lunar Lander ML Agent
Three ways to land the Gymnasium LunarLander-v3 craft, compared against each other: a hand-written rule-based controller, a decision tree trained to imitate recorded keyboard and agent play, and Q-learning over a discretised state space. The observations are continuous and Q-learning needs discrete states, so the crux is discretisation: manual binning by domain knowledge is measured against k-means vector quantisation swept across 32 to 256 clusters.
04. Skills
What I work with
Languages
- Python
- TypeScript
- JavaScript
- Java
- C/C++
- R
- SQL
- NoSQL (MongoDB, Redis)
AI/ML
- RAG Pipelines
- Vector Search & Embeddings
- Multi-Agent Orchestration
- LLM Fine-Tuning
- LangChain
- LangGraph
- LangSmith
- Braintrust
- Scikit-learn
Frameworks
- React
- Node.js
- Flask
- FastAPI
- Spring Boot
- Pydantic
- NumPy
- Pandas
Tools
- AWS (Bedrock, S3)
- Docker
- Kubernetes
- Kafka
- Git
- Linux
- PostgreSQL
- DynamoDB
- ElasticSearch
- Terraform
- CI/CD (GitHub Actions)
- Vercel
- GraphQL
- Playwright
- Snowflake
Concepts
- LLM-as-a-Judge
- Agentic Workflows
- Prompt Engineering
- Mechanism Design
- Auction Theory (VCG, Myerson)
- Yield Optimization
- REST APIs
- Microservices
05. Education
Academic background
University of Maryland, College Park
Aug 2023 – May 2027
BS Computer Science
Machine Learning Concentration · Minor in Business Analytics
GPA: 3.67- ▸Teaching Assistant, CMSC131 Object-Oriented Programming I
- ▸Research at Abellon Clean Energy: which waste feedstock properties predict efficient Waste-to-Energy conversion, over 100,000+ rows of industrial sensor data. Deployed to operations, +23% productivity.
Universidad Carlos III de Madrid
Jan 2026 – May 2026
Semester Study Abroad
Coursework: Machine Learning I, Artificial Intelligence
Madrid, SpainCertifications
- Prompt Engineering for Developers, DeepLearning.AI
- Google Python Programming Certificate
- Google Advanced Data Analytics Certificate