I'm Eesh Saxena, a CSE student at IIIT Senapati, Manipur focused on DSA, Full Stack Development, AI/ML, and Research.
- Location: Gandhinagar, Gujarat, India
- Email: eeshsaxena@gmail.com
- LinkedIn: linkedin.com/in/eeshsaxena
- Currently working on: BolKota (Rajasthani voice assistant), Cardiac Edge AI MCU deployment
- Currently learning: Graph RAG, TinyML, Low-resource NLP
- Open to research internships and collaborations
Under Dr. Abhisek Paul (Onsite)
- Analyzed the framework from Zhang (IEEE SPL 2011) on Reversible Data Hiding in Encrypted Images, focusing on secure embedding with exact image recovery.
- Studied and experimentally evaluated techniques from 30+ related research works to compare embedding strategies and reversibility guarantees.
- Developed and tested the encryption-data embedding-extraction process ensuring separability between data retrieval and image reconstruction.
- Measured embedding capacity and reconstruction fidelity using PSNR and embedding rate metrics, contributing toward ongoing publication-oriented work.
|
First open, annotated Rajasthani-Hindi code-switched NLP corpus — 50,000 sentences from Twitter/X and ShareChat. Fine-tuned MuRIL models outperform GPT-4o on sentiment, NER, and toxicity detection. Extends to: BolKota (voice assistant) — the corpus pipeline feeds directly into ASR and NLU components for a Rajasthani voice assistant. |
5-class cardiac arrhythmia detection on Arduino Nano 33 BLE. Novel spectral knowledge distillation loss (L_spectral) — 140x compression (1.27M to 9,100 params), 23.4 KB INT8, ~99% F1 with ECG+PPG fusion. Target Q2 journal. |
|
Cloud-native distributed file storage with 8MB chunking, 3-way replication, SHA-256 deduplication, AES-256-GCM encryption, resumable uploads, CDN delivery, and file versioning. Dockerized with PostgreSQL + Redis. |
Graph-based RAG pipeline converting unstructured text into entity-relation triples. Neo4j knowledge graph with query-aware Ref(p) scoring and entropy-based conflict detection (DeltaH) to reduce factual inconsistencies over baseline vector RAG. |
- Codeforces Specialist - Highest Rating 1582 - eeshsaxena (Top 1% in India)
- CodeChef 4-Star - Highest Rating 1866 - kidkrish (Top 1% in India)
- LeetCode Guardian - Highest Rating 1873 - eeshsaxena (Top 5% worldwide)
- Solved 1,500+ problems across 50+ contests on all platforms
- Global Rank 1 in CodeChef Starters 180 (Div. 3) among 10,000 participants
- Global Rank 75 in LeetCode Weekly Contest 446 among 27,686 participants
- Global Rank 78 in CodeChef Starters 183 (Div. 2) among 10,000 participants
| Platform | Profile | Rank & Rating | Top % | Problems Solved | Contests |
|---|---|---|---|---|---|
| LeetCode | eeshsaxena |  |
Top 5% (Global) | 800+ | 25+ |
| Codeforces | eeshsaxena |   |
Top 1% (IN) | 300+ | 15+ |
| CodeChef | kidkrish |  |
Top 1% (IN) | 400+ | 15+ |
-1500%2B-brightgreen?style=for-the-badge){kind=icon}
| Category | Technologies |
|---|---|
| Languages | C++, Python, JavaScript, TypeScript, PHP, SQL |
| Frameworks & Libraries | React.js, Node.js, Next.js, Express.js, PyTorch, NumPy, Pandas |
| Databases | MySQL, MongoDB, Supabase, Firebase |
| Tools & Platforms | Git, Docker, Linux, AWS, Postman, Vercel |
Production-ready code forged in dragonfire.
"Code like a developer, think like an architect, and build like a dragon."