Skip to content
View SeivenBell's full-sized avatar
🎯
Focusing
🎯
Focusing
9 followers · 10 following

Achievements

Achievement: Pull SharkAchievement: YOLO

Achievements

Achievement: Pull SharkAchievement: YOLO

Organizations

@OpenQuantumDesign

Block or report SeivenBell

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

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

Report abuse
SeivenBell/README.md

🚀 Severyn Balaniuk | ML / AI Engineer and Founder

linkedin

I specialize in building ML / AI systems, particularly for deep tech products. My experience spans distributed training pipelines, compiler optimizations, and semi supervised learning.

🔨 Tech Stack

Frameworks

Cloud and DevOps

📌 Featured Projects

Project Description
Quantum Architecture Search Harness ⚛️ Cirq + Gymnasium + OpenRouter agent harness for QAS reward-shaping. Multi-seed sweeps caught an underpowered entanglement-reward result and falsified it under stronger evaluation.
RL Environment for VQE Circuit Optimization 🧠 Gymnasium/PyTorch env with goal-conditioned replay buffers, 2.84× wall-time speedup & 65% better curriculum alignment vs DDQN baselines.
Multi-Agent Ops Anomaly Detection 🏗️ Auditor/compliance/schedule/synthesizer agent pipeline (Azure OpenAI, AI Search, Blob, SQLite) over 5,316 synthetic construction records, raising anomaly recall 13% → 80%.
Quantum Resource Estimation Demo 📊 Gradio + FastAPI service on GCP Cloud Run with Auth0, giving research stakeholders direct hardware-target scaling visibility without API access.
Semi-Supervised Fluorescence Classification 🔬 Distributed PyTorch/CUDA pipeline over 500K images, +31% top-1 accuracy while cutting labeled-data needs 80%.
Wildfire Growth Prediction 🛰️ CNN model predicting wildfire spread at 89% accuracy from satellite imagery + weather time series. Winner, Wildfire Hackathon 2024.

Pinned Loading

  1. Wildfire_Hackathon_CNN Wildfire_Hackathon_CNN Public

    Predicting wildfire growth using complex time series geospatial and weather datasets with advanced CNN achitechture.

    Jupyter Notebook

  2. AgentFlowCorp/Front-End AgentFlowCorp/Front-End Public

    Client-based route optimization app for real estate agents. Status: shelved MVP, built solo in ~1 week (Nov 2025) to validate the idea. Paired with the AgentFlow Backend. Not actively maintained.

    Vue

  3. cocktail_advisor cocktail_advisor Public

    A Python-based chat application that integrates with a large language model (LLM) to create a Retrieval-Augmented Generation (RAG) system using a vector database for cocktail recommendations and in…

    Python

  4. ChatGPT2_Paper_Reproduction ChatGPT2_Paper_Reproduction Public

    Reproducing paper on ChatGPT 2 with possible improvements and additional documentation

    Jupyter Notebook

  5. Machine_Learning_for_Physics_Problems Machine_Learning_for_Physics_Problems Public

    Developing various models such as Feedforward Neural Networks, Recurrent Neural Networks, Variational Autoencoders, and Boltzmann Machines.

    Python 1

  6. Data_Science_and_Statistical_Learning_Projects Data_Science_and_Statistical_Learning_Projects Public

    Implemented and used image processing, feature extraction, clustering, decision trees, random forest, SVR, SVM, model evaluation (RMSE, precision, recall), Scikit-learn, Pandas, NumPy, OpenCV, PIL.

    Jupyter Notebook