Akash Dubey

Akash Dubey

I've been working in software for over 10 years 😉

Akash Dubey

Akash Dubey

Engineer | Computer Science + Mathematics @ Rutgers University Honors College Location: Berkeley Heights, New Jersey, United States Contact: akashdub06@gmail.com | 212-814-3681 Links: Portfolio | GitHub

Summary

I’ve been working with software and mathematics for over 10 years. Enthusiastic about AI/ML, Agentic Systems, Optimization, and High-Performance Computing.

Experience

Rutgers Economics Labs (Oct 2024 - Present)

President (Jan 2026 - Present)

  • In charge of all research operations and expanding the organization.
  • Leading a team developing the Rutgers Agentic Intelligence Lab: An internal project run by Rutgers Economics Labs (REL) aimed at creating a general-purpose framework for economic analysis. The framework makes it easy to clean, analyze, and connect data sources in an ontology-like format, enabling researchers to build structured, reusable pipelines for complex economic datasets.
  • Leading quantitative research projects for partners including NJDOL, NJDEP, NJBPU, Federal HUD, U.S. Census Bureau; applied econometric models in Python/R to analyze large datasets and project economic trends.

Research Director (July 2025 - Feb 2026)

  • In charge of overseeing and sourcing all research projects. Designed studies for NJDOL and Virginia Consensus for Higher Education in Prison.

Team Lead (Feb 2025 - Aug 2025)

  • Led a team of five dedicated student researchers to analyze the Urban Enterprise Zone program for the New Jersey Department of Community Affairs. Performed comprehensive analysis of census data with econometric methods in Python and R.

Economic Researcher (Oct 2024 - Feb 2025)

  • Analyzed NJ Clean Energy Programs, EV incentives. Synthesized 20+ years of historical data from various sources to extract patterns and create projections of economic funding variables. Scraped every historic NJ state budget PDF.

Scarlet Sync (Jan 2025 - Present)

Founder

  • Join 750+ users planning their schedule from thousands of classes and 500+ degree options.
  • Created a modern Rutgers class scheduling app with thousands of users for degree planning and notifications, utilizing AI-assisted parsing of unstructured data from legacy university services.
  • Architected a robust data ingestion pipeline using Python and LLMs to reverse-engineer legacy data, indexing over 2,500 classes and 500+ degree programs for real-time querying.
  • Negotiating enterprise deal with Rutgers IT to replace outdated systems with AI-enhanced tools.

Lykke (Sep 2025 - Present)

Software Engineer

  • Helping design and expand a multi-model AI pipeline to ingest, parse, and structure high-velocity, multi-modal data (texts, voice memos, docs) for a conversational RAG-based and agentic chat application.
  • Engineered a fully local, browser-based RAG system by implementing a local embedding model optimized with WebGPU for client-side hardware acceleration, ensuring high performance and privacy.
  • Architecting the integration of Canvas LMS to build a rich, domain-specific knowledge graph, enhancing AI reasoning and significantly reducing hallucinations in structured outputs for students.

Rutgers Department of Computer Science (Jan 2026 - Present)

Research Assistant (Advisor: Prof. Zhao Zhang)

  • Exploring second-order optimizers (e.g., Shampoo) to accelerate convergence in large-scale ML training.
  • Designing agentic systems for non-sequential code generation, focusing on decoupling reasoning from syntax.

Algorithmic Robotics and Control Lab - Rutgers CS (Jan 2026 - Present)

Research Assistant (Advisor: Prof. Jingjin Yu)

  • Optimizing robot motion planning algorithms using CUDA for parallel computation in high-dimensional spaces (pRRTC and FCIT*).
  • Working with foundational vision language action models (NVIDIA GR00T).

Kwan Lab, Rutgers Cancer Institute of New Jersey (Jan 2026 - Present)

Computational Biology Research Assistant (Advisor: Prof. Kelvin Kwan)

  • Engineered an AlphaGenome AI-powered deletion sensitivity scanner to quantify TAD boundary functional importance.
  • Built a custom 3D polymer simulation engine using overdamped Langevin dynamics to reconstruct chromatin architecture from Hi-C contact matrices.

Samaritan Scout (May 2023 - Aug 2025)

Full Stack Engineer (May 2025 - Aug 2025)

  • Developed an agentic scraping system leveraging OpenAI and Gemini structured outputs to autonomously extract data from over 100,000 non-profit websites.
  • Developed a full-stack search engine using React, TypeScript, and PostgreSQL to index 35,000+ volunteer opportunities, implementing semantic search algorithms.

Frontend Software Engineer (May 2023 - Aug 2024)

  • Learned React, TypeScript, and MaterialUI to help create the home, about, and donate page of website.

Education

Rutgers University Honors College, School of Arts and Sciences (Anticipated May 2027) Bachelor of Science in Computer Science and Mathematics

  • GPA: 3.97/4.0
  • Relevant Coursework: Tensor Networks, Systems Programming, Algorithms, Data Structures, Statistics, Honors Calculus III & IV, Math Theory of Probability, Computer Architecture.

Projects

  • Local Coworker: Native macOS Computer-Use Agent: Built a native desktop agent capable of autonomous computer use by integrating Gemini Vision with macOS Accessibility APIs, AppleScript, and JXA. Implemented the Model Context Protocol (MCP).
  • SLM Optimization & Post-Training (GSM8k): Fine-tuned small language models on the GSM8k math reasoning dataset. Deployed models on the Rutgers High Performance Computing (Amarel) cluster.
  • Crazy All-in One AI Assistant: A comprehensive AI assistant with specialized agents, advanced memory systems, and seamless Google services integration.
  • TopCoder Reimbursement Challenge (Top Performer): Reverse-engineered a 60-year-old undocumented travel reimbursement system from 1,000 I/O examples using XGBoost and feature engineering — recognized as a top performer and invited to interview at 8090 AI.
  • HiC-TAD-Library — 3D Genome Analysis & AI-Powered Enhancer Design: A Python toolkit for analyzing 3D chromatin architecture from Hi-C data, with AlphaGenome AI integration for in-silico structural variant prediction and synthetic enhancer design.
  • Grokipedia API: A comprehensive Python and JavaScript/TypeScript client library for accessing Grokipedia’s vast knowledge base.
  • Lykke — AI Study Platform: An AI-powered study platform connecting Canvas LMS, notes, and course materials to generate flashcards, quizzes, infographics, and personalized study plans.
  • Scarlet Sync: A complete scheduling and degree planning solution for Rutgers.
  • NOSTRADAMUS — NJ Geospatial Intelligence Platform: A 3D geospatial simulation platform for New Jersey with live flight tracking, drone flythrough, and layered economic & infrastructure data.
  • Data Cleaning Assistant: An AI agent that takes a CSV and plain-English instructions and returns cleaned data plus the Python code that did it.
  • A* Path Finder Visualizer: A pygame program to visualize the A* algorithm in action.
  • AP Calculus Problem Generator: A calculus problem generator made for a final project.
  • ASD Racing Game: A 3D racing game made with the Godot game engine.
  • American Statistical Association Fall Data Challenge 2023: Analysis of the Stack Overflow Developer Survey data.
  • ESL Worksheet Generator: A Python App that generates basic English worksheets.
  • Economic Series Plotter: A plotter for FED and BLS data.
  • FBLA Coding and Programming Project: CTE Partnerships, a website bringing information about local businesses together!
  • FBLA Portfolio Making Android App: An Android app for people to make and share portfolios.
  • FRC Robot Code: Team 1257’s 2024 Robot Code.
  • High School Programming Projects: MSDS and Advanced Software Development Projects.
  • MSDS Final: Depth Estimation: Depth Estimation Final Project.
  • Rutgers Bus Analysis: A comprehensive data analysis of the Rutgers University bus transit system.
  • The Campus Chronicle Website: The UCVTS Newspaper Club’s new website.
  • q-Binomial Theorem Presentation - Spring 2025 DRP Program: Research and presentation of the q-Binomial theorem and symmetric functions for the 2025 Rutgers Directed Reading Program.
  • FlaminGo! (HealthHack 2025): An iOS app empowering women 50+ to build stronger bones and prevent falls. Leverages the Vision framework for real-time camera-based posture/balance analysis and HealthKit for movement tracking, calculating a comprehensive fall risk score completely on-device.
  • Economic Drivers of Puerto Rican Migration: A data-driven interactive web experience analyzing Puerto Rican migration patterns from 1950 to 2024. Fused quantitative datasets (U.S. Census, IPUMS, FRED) with literary analysis, utilizing Python/Plotly for complex interactive visualizations and a custom HTML/CSS scroll-driven frontend.

Skills & Languages

  • Programming Languages: Python, C++, CUDA, Java, SQL, TypeScript, R, Bash, Git.
  • Technologies/Frameworks: PyTorch, WebGPU, Transformers, Generative AI, Pandas, AWS, Scikit-learn, HPC Clusters.
  • Languages: English (Native or Bilingual), Spanish (Elementary), Hindi (Limited Working).

Certifications & Awards

  • Awards: 8090 AI Top Coder (5th), Rutgers Shark Tank (2nd and 3rd), ASA Fall Data Challenge (Top 3), Dean’s List Semifinalist, High Honor Roll, RIT High School Science and Math Award.
  • Certifications: Microsoft PowerPoint (Microsoft 365 Apps and Office 2019), CompTIA IT Fundamentals (ITF+), Java Foundations Certified Junior Associate, IC3 Digital Literacy Certification GS6 Master.