Skip to content
View Ultron09's full-sized avatar
πŸ—Ώ
Creating
πŸ—Ώ
Creating

Organizations

@AirBorneAI

Block or report Ultron09

Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
Ultron09/README.md

Typing SVG

LinkedIn Email GitHub Discord


🎯 EXECUTIVE SUMMARY & RECRUITER HIGHLIGHTS

AI Researcher & Systems Software Engineer building first-principles deep learning architectures, intelligent AI proxies, multimodal systems, and high-concurrency microservices. Experienced in bridging theoretical AI research (NeurIPS evaluation frameworks, continual learning, dual-system neural networks) with battle-tested production software engineering (C++, Python, PyTorch, Node.js, Redis, Supabase, Docker).

πŸ’Ό Target Positions

  • πŸ§ͺ AI Scientist / Research Engineer β€” Deep Learning Architectures, Continual Learning, Dual-System Neural Nets, Transformer & CNN Math from Scratch.
  • πŸ“Š Data Scientist / Analytics Engineer β€” Predictive Modeling, Statistical Inference, Exploratory Data Analysis, SQL Pipelines.
  • ⚑ AI Engineer / GenAI Specialist β€” LLM Infrastructure & Proxies, RAG, Agentic AI Workflows, Natural Language Voice Processing.
  • πŸ’» Software Development Engineer (SDE / Backend / Mobile) β€” Clean Architecture, High-Throughput C++/Python/Node Services, Redis Caching, Cross-Platform Mobile Apps.

πŸ› οΈ CORE COMPETENCIES MATRIX

🧠 AI Scientist

  • Deep Learning & Neural Architectures: Continual Learning, Orthogonal Gradient Descent (OGD), Random Gated Weights (RGW), Transformers, CNNs.
  • First-Principles Mathematics: Backpropagation, custom loss derivations (Focal, Huber, Cross-Entropy), GELU/Swish activation math.
  • Research & Evaluation: NeurIPS benchmark suites, Backward Transfer (BWT), Forward Transfer (FWT), W&B telemetry.
  • Frameworks: PyTorch, NumPy (pure math implementation), TensorFlow, OpenCV.

πŸ“ˆ Data Scientist

  • Statistical Analysis & Modeling: Hypothesis Testing, Regression, Classification, Clustering, Predictive Analytics.
  • Exploratory Data Analysis: Feature Engineering, Data Cleaning, Correlation Analysis, Statistical Distribution Modeling.
  • Data Infrastructure & Visuals: Pandas, NumPy, SciPy, Plotly, Seaborn, Matplotlib.
  • SQL & Data Management: Complex SQL Queries, Relational Database Schemas, Data Aggregation.

⚑ AI Engineer

  • LLM Integrations & AI Proxies: Anthropic Claude 3.5 API, NLP extractions, prompt orchestration, automated AI strategy synthesis.
  • Multimodal Processing: Audio/Voice log parsing (`.m4a`), natural language macro extraction, structured JSON serialization.
  • RAG & Vector Search: Context-aware retrieval, semantic search, Agentic workflows.
  • MLOps & Deployment: Model Serving, Docker containerization, REST API design.

πŸ’» Software Development Engineer (SDE)

  • Core Languages: Python, C++, C, JavaScript, TypeScript, SQL, Bash.
  • High-Performance Backend: Node.js, Express.js, RESTful APIs, In-Memory Redis ($O(\log N)$ sorted set leaderboards).
  • Database & Cloud Infrastructure: Supabase, PostgreSQL, MySQL, Docker, Git.
  • Mobile & Cross-Platform: React Native, Expo, native Android/iOS integration.


⚑ AUTHENTIC & VERIFIED FEATURED PROJECTS

Project & Role Focus Authentic Technical Highlights Links & Tech Stack
🧠 ANTARA - NeurIPS Evaluation Suite
Target: AI Scientist / Deep Learning Researcher
β€’ Mathematically hardened benchmark suite evaluating dual-system neural architectures (System 1: OGD, System 2: RGW).
β€’ Prevents catastrophic forgetting on 10-task CIFAR-100 partitions without raw data rehearsal.
β€’ Multi-node distributed cluster evaluation with live Weights & Biases (W&B) telemetry tracking BWT, FWT, and step latency.
[GitHub Repo]

PyTorch Python
CUDA W&B
Docker
πŸ€– GPT from First Principles (Pure NumPy)
Target: AI Scientist / AI Engineer / SDE
β€’ Complete from-scratch implementation of Generative Pre-trained Transformer (GPT) using strictly NumPy.
β€’ Full forward & backward propagation derived from scratch without deep learning frameworks.
β€’ Built multi-head attention, layer norm, positional embeddings, causal masking, and Adam optimization math.
[GitHub Repo]

Python NumPy
Linear Algebra
Backprop
πŸ‘οΈ NumPy CNN Engine from Scratch
Target: AI Scientist / Computer Vision Engineer
β€’ Pure NumPy Convolutional Neural Network engine with zero high-level framework dependencies.
β€’ Manual backprop derivation for Conv2D, MaxPool, AvgPool, and BatchNorm2D.
β€’ Implemented 8 activations (GELU, Swish), 7 losses (Focal, Huber), and 6 optimizers (Adam, Nadam, SGD momentum).
[GitHub Repo]

Python NumPy
Math & Backprop
Optimization
⚑ A.R.C. Intelligent Performance Proxy
Target: AI Engineer / SDE Backend Developer
β€’ Express.js command center and AI proxy powered by Claude 3.5 Sonnet.
β€’ Multi-endpoint pipeline handling natural language voice log parsing (Multer `.m4a`), macro extraction, and daily strategy synthesis.
β€’ In-memory Redis store managing $O(\log N)$ sorted set leaderboards synced to Supabase.
[GitHub Repo]

Node.js Express
Claude API Redis
Supabase
πŸ“± Skyforce HR & OnSpace Mobile Suite
Target: SDE / Cross-Platform Mobile Engineer
β€’ Enterprise cross-platform mobile application built with React Native, Expo, and TypeScript.
β€’ Seamless backend integration with Supabase for authentication, real-time data sync, and relational queries.
β€’ Native Android/iOS deployment pipelines with smooth Lottie micro-animations and calendar workflows.
[GitHub Repo]

React Native Expo
TypeScript Supabase


πŸ›‘οΈ TECHNOLOGY COMMAND CENTER

πŸ’» Languages & Systems

Python C++ C SQL TypeScript JavaScript CUDA Bash

🧠 AI, Machine Learning & Research

PyTorch TensorFlow NumPy WandB OpenCV

πŸ“Š Data Science & Analytics

scikit-learn Pandas SciPy Plotly Matplotlib Seaborn

πŸš€ Web Backend, Mobile & Infrastructure

Express.js Node.js React Native Expo Redis Supabase Docker


πŸ’» INTERACTIVE DEVELOPER CONSOLE

#!/usr/bin/env python3
"""
Suryaansh Prithvijit Singh - Verified Engineer Profile
Target Roles: AI Scientist | Data Scientist | AI Engineer | SDE
"""

class SuryaanshProfile:
    def __init__(self):
        self.name = "Suryaansh Prithvijit Singh"
        self.github = "https://github.com/Ultron09"
        self.email = "connect.singha@gmail.com"
        self.linkedin = "https://www.linkedin.com/in/suryaansh-prithvijit-singh/"
        
        self.active_research = [
            "ANTARA: Continual Learning & Dual System OGD/RGW Architectures (NeurIPS)",
            "From-Scratch Autoregressive Transformers (NumPy-GPT)",
            "Pure NumPy CNN Layer & Backprop Mathematics Engine"
        ]
        
        self.production_systems = [
            "A.R.C.: Express & Claude 3.5 Performance Proxy with Redis Sorted Sets",
            "Skyforce HR: React Native & Expo Cross-Platform Agentic Mobile Suite"
        ]

    def status(self) -> dict:
        return {
            "hire_status": "OPEN_TO_ELITE_ROLES πŸš€",
            "primary_focus": ["AI Scientist", "Data Scientist", "AI Engineer", "Software Engineer (SDE)"],
            "availability": "Immediate / Global Remote & Relocation"
        }

if __name__ == "__main__":
    dev = SuryaanshProfile()
    print(f"⚑ {dev.name} [{dev.github}] is active.")
    print(f"🎯 Status: {dev.status()['hire_status']}")


πŸ“Š ANALYTICS & GITHUB ACTIVITY DASHBOARD

GitHub Stats GitHub Streak
Top Languages Activity Graph


🀝 CONTACT & RECRUITER CONNECT

πŸ’Ό Recruiting or Hiring?

Actively interviewing for AI Scientist, Data Scientist, AI Engineer, and Software Development Engineer (SDE) roles.

⚑ Expected Response Time: Within 12-24 Hours

Pinned Loading

  1. BrainTumor BrainTumor Public

    Detecting and Classifying Brain Tumor by Creating a CNN model based on VGG16

    Jupyter Notebook 1

  2. Numpy-Cnn Numpy-Cnn Public

    Advanced CNN Training from Scratch with NumPy. A modular, educational Deep Learning framework implementing Convolutional Neural Networks without high-level libraries. Features 15+ layer types, 8 ac…

    Python 2 1

  3. Numpy-Transformer Numpy-Transformer Public

    A pure NumPy implementation of GPT built from scratch for educational purposes. Features manual backpropagation, multi-head attention, and a custom Adam optimizer with zero deep learning frameworks.

    Python 2

  4. LeetCode LeetCode Public

    Practice and Documentation

    C++

  5. huggingface/transformers huggingface/transformers Public

    πŸ€— Transformers: the model-definition framework for state-of-the-art machine learning models in text, vision, audio, and multimodal models, for both inference and training.

    Python 165k 34.4k

  6. sqliteai/warp sqliteai/warp Public

    Run the full 2.78-trillion-parameter Kimi K3 model or GLM-5.3-Flash beyond available RAM by streaming activated weights directly from NVMe. A dependency-free, embeddable C inference engine.

    C 2.3k 172