I'm a Computer Science student at Queens College, CUNY graduating in May 2026, with deep expertise in AI/ML Engineering and Full-Stack Development. My work sits at the intersection of rigorous mathematical foundations and practical system design — I don't just use AI tools, I build them from first principles.
Currently, I'm the founder of WORT.AI, an autonomous research agent that outperforms GPT-5.1 benchmarks on complex query generation. My approach to AI engineering is rooted in understanding transformer architectures at the mathematical level, not just API calls.
Location → Jersey City, NJ
Focus → AI Agents • LLM Systems • Production ML • Backend Architecture
Philosophy → Build systems that think, not just respond
|
Aug 2024 – Dec 2025 · Full-time Built the central workforce platform for 5,000+ caregivers — not a side project, the actual production system handling shift scheduling, real-time payroll, and performance tracking. Key Deliverables:
Stack: |
|
Sep 2024 – Jan 2025 · Research · Prof. Chialing Tsai Led a research team investigating LLMs for tabular data processing — a domain where traditional ML still outperforms most LLM approaches. Results:
|
|
Jun 2023 – Aug 2023 · India Developed REST APIs for educational software serving 10,000+ users across India.
|
WORT is not another ChatGPT wrapper. It's a recursive research agent that uses BFS algorithms to autonomously detect information gaps and navigate complex web sub-graphs, producing comprehensive reports with full citations.
Technical Architecture:
| Component | Implementation |
|---|---|
| Memory System | Custom hybrid combining Qdrant (Dense + Sparse BM25) with Cross-Encoders and RRF re-ranking |
| Query Generation | RLVR fine-tuned Qwen models — elevated accuracy from 71% → 94%, surpassing GPT-5.1 (91%) |
| Data Collection | Parallel scrapers navigating dynamic, human-interaction-heavy websites |
| Orchestration | LangGraph state machine with conditional routing and human-in-the-loop checkpoints |
Stack: LangGraph Qdrant RLVR Qwen Cross-Encoders FastAPI Custom Scrapers
A productivity platform that goes beyond time tracking. ION categorizes work into HyperFocus, SharedWork, and CoreWork modes, letting users audit and optimize their cognitive output patterns.
- Teamspaces: Aggregate individual consistency metrics into group dashboards — visibility without micromanagement
- Real-time Sync: Supabase subscriptions keeping goal tracking consistent across all devices
- Friction-free Logging: Designed for daily use, not weekly guilt trips
Stack: TypeScript React Supabase Vite Advanced Analytics
-2A5C42?style=flat-square){kind=spacer}
Bachelor of Science in Computer Science · Sept 2021 – May 2026
Relevant Coursework:
| Mathematics | Machine Learning & AI |
|---|---|
| Advanced Calculus | Natural Language Processing |
| Linear Algebra & Vectors | Computer Vision |
| Abstract Algebra | Machine Learning |
| Game Theory | Deep Learning |
- Top 5 Runner-up at HackRU 2024 (Rutgers University Hackathon)
- 300+ LeetCode problems solved — focus on backtracking, dynamic programming, and complex graph algorithms
- Certifications: Machine Learning A-Z (Python) · Natural Language Processing (DeepLearning.ai)
Tech Stack
