I'm a Computer Science undergraduate at NJIT's Ying Wu College of Computing, co-founder of Contextral, and the sole CS and firmware engineer on NASA's RockSat-X A.R.R.O.W. mission.
Most of my work sits where AI meets constraint: air-gapped labs, compliance boundaries, municipal archives, and a payload that had to survive a suborbital flight.




A company, a NASA payload, a research lab, a community, and a fellowship, ordered by what carries the most weight, not by date.

Building on-premise AI infrastructure for regulated institutions, covered by NJIT News. Flagship product Virtual Archives is deployed at the Newark City Clerk's Office: HIPAA/FERPA-compliant document intelligence via Ollama, Qdrant, and LangChain.
Sole CS and firmware engineer for a suborbital payload. Wrote all flight code for SCOUT (SAMD21, Langmuir probe + XBee transmitter) and Overwatch (RA4M1, IMU logger + receiver) through CoDR, PDR, and CDR. Flew the payload in June 2026, then led post-flight telemetry analysis: processed 49,171 IMU samples, diagnosed a 75.3% RF reception rate, and authored the mission's final technical report.

Oversee all lab experiments and run gesture-based XR studies using Oculus, Unity, Azure, and C#. Nearly doubled participant engagement with 100% consent form completion. Led an independent sub-study in the VROOM project, published at ACM IMX 2026.

Selected from a competitive applicant pool to advise the YWCC Advising Team on the student experience across computer science, data science, and informatics.

A Newark-rooted startup community and event series. Ecosystem building rather than engineering: convening founders, operators, and students around the city's growing tech scene.

A competitive, named fellowship backing the founder track with institutional support. Alongside NVIDIA Inception membership for Contextral and Dean's List standing at NJIT.
Also: LLM/NLP Engineering Research Assistant at NJIT's Dept. of Humanities & Social Sciences (under Prof. Michael Laudenbach) · Chief Product Officer at Project Elevate · Base44 Ambassador for the NJIT campus · Mentor & Program Experience Advisor at CodePath · Software Engineer at Develop for Good · E-board Social Media Manager at the NJIT Center for Entrepreneurship.
Retrieval on hardware nobody wants to replace, agent behaviour under corpus control, and collaboration inside XR.
Invited as an undergraduate to review for IEEE TKDE. Currently writing on local LLM deployment behaviour across open-source models with NJIT's Dept. of Humanities and Social Sciences.
Hackathon builds, products shipped at Contextral and Project Elevate, and a contribution to GPU tooling.
Four-agent architecture: Specialization, VersionPatrol, QueryDecomposition, and Escalation. Passed 12 of 12 synthesis tuning benchmarks. Built for the Qwen Global AI Hackathon, $70K prize pool.
Draw a platformer on paper, photograph it, play it in the browser. Hand-built AABB physics engine with coyote time, jump buffering, and AI autoplay. 1st Place Amateur Track and 1st Runner-Up, AI + Hardware at HackPrinceton.
Real-time AI translation with mixed-input rooms across voice, text, and SMS/WhatsApp, routed through Claude. Experimental ASL recognition via vision API, plus an ESP32 hardware companion. Built at the NJIT Hackathon.
Open source contributor to a 2.4k-star project. Opened Issue #47 proposing a layered GPU metadata resolver, then merged PR #75 adding Kepler-era Quadro entries with Vulkan-only tagging and pytest coverage.
HIPAA/FERPA-compliant document intelligence running entirely on-premise, built on Ollama, Qdrant, and LangChain. Contextral's flagship product, deployed at the Newark City Clerk's Office.
YOLOv8 computer vision system built with the Newark Department of Public Works, cutting response time by 40%. Paired with an OCR and AI pipeline that digitised decades of city archives.