LOISEKKยฎ

Initializing neural runtime0%

Selected Projects

AI / ML ENGINEER ยท YASH BRAHMANKAR

Open Source View Resume โ†—

Yash Brahmankar
AI / ML Engineer & Open-Source Developer
Founder & Lead Developer, OpenML

"I build intelligent systems,
developer tools & AI-powered experiences โ€”
from ML pipelines to autonomous agents."

02/

The Founder's Vision

Architecting OpenML โ€” a local-first ecosystem of visual machine learning IDEs and deterministic DAG execution engines built for production engineering.

03/

Engineering in Motion

From production MLOps pipelines to 3D autonomous agents and telemetry, I build at the intersection of AI, high-performance systems, and open source.

01/

I build intelligent OpenML AutoML Terminal systems that move beyond the notebook. From machine learning pipelines to AtelierOS 3D Spatial Agent Workspace autonomous agents, I turn ideas into resilient, working software. At the intersection of AI, engineering, and interactive experiences.

Selected Projects
14
Model Benchmarks
360+
Pipeline Reliability
99%
Core Philosophy
01

02/

Fourteen engineered systems and research breakthroughs orbiting the core stack.
Inspect any mission to view technical architecture.

Filter by architecture

    Scroll to rotate project orbit ยท Hover a card to clear glass & reveal telemetry

      03/

      Built in the
      Open Source

      2026Active GitHub@loisekk 100%Open

      DETERMINISTIC EXECUTION LIFECYCLE

      How OpenML Eliminates Silent Failures

      Five stages, one continuous pipeline. Follow the cryptographic execution path โ€” or select any node to inspect the underlying engineering.

      STAGE 01 / 05

      Data Contracts โ€” Cryptographic Schema Validation

      Every dataset begins with strict cryptographic verification. Pydantic schema contracts enforce type safety, and raw data blobs are hashed with SHA-256 before any transformation begins โ€” guaranteeing zero silent drift between training and inference.

      โ€ข SHA-256 STATE HASHING โ€ข STRICT PYDANTIC CONTRACTS โ€ข ZERO SILENT DRIFT

      OpenML Initiative ยท Founder & Lead Developer

      Engineering Deterministic Systems for the AI Frontier

      As founder and lead developer of OpenML, I design developer tools and intelligent software that replace fragile scripts with verifiable pipelines. From local-first machine learning IDEs to 3D autonomous agent swarms, my goal is to make advanced AI workflows accessible, deterministic, and 100% open source.

      04/

      Deep Learning & Vision Stance Production MLOps Stance Autonomous Agents Stance Telemetry & 3D Systems Stance
      Live Telemetry FOCUS: 01 // DEEP LEARNING & VISION

      Architecting neural networks, fine-tuning vision and sequence backbones, and evaluating high-dimensional loss landscapes. Deep specialization in PyTorch, CNNs, transformer architectures, and explainable AI.

      • PyTorch & CUDA
      • Vision Transformers
      • Loss Landscapes
      • Explainable AI
      TECHNICAL SKILLS BY DOMAIN // VERIFIED REPOSITORIES 42 TECHNOLOGIES ACTIVE

      Engineered Stack & Specializations

      Grouped by engineering practice, verified against active repositories across the OpenML ecosystem, autonomous agent frameworks, and high-frequency telemetry systems.

      ๐Ÿง  01 // NEURAL CORE
      ็Ÿฅ

      AI / Machine Learning / Deep Learning

      Deep neural network architectures, custom CNN backbones, and GPU-accelerated inference pipelines.

      PyTorch TensorFlow Scikit-learn CNNs Vision Transformers XGBoost ONNX Runtime CUDA
      VERIFIED ON: THERMOWATCH ยท F1 ML ยท BASECAMP
      ๐Ÿค– 02 // AGENTIC SWARMS
      ่ก“

      LLMs / Agentic Systems & RAG

      Autonomous multi-agent orchestration, stateful DAG routing, and deterministic tool verification.

      LangChain LangGraph Groq Whisper LlamaIndex Multi-Agent Swarms Tool Verification RAG Pipelines
      VERIFIED ON: ATELIEROS ยท OPENHIVE ยท BRAIN NEXUS
      ๐Ÿ 03 // DATA FOUNDRY
      ็ฎ—

      Python / Data Science & Analytics

      Geospatial telemetry, multi-terabyte raster analytics, statistical modeling, and interactive dashboards.

      Python 3.12 NumPy Pandas GeoPandas GDAL Plotly Streamlit Statistical EDA
      VERIFIED ON: THERMOWATCH ยท F1 TELEMETRY ยท GEO-PIPELINE
      โšก 04 // LOW-LATENCY CORE
      ๆ ธ

      Backend Engineering & APIs

      High-frequency WebSocket event brokers, asynchronous microservices, and cryptographically verified APIs.

      FastAPI REST APIs WebSockets Asyncio Pydantic Microservices Node.js
      VERIFIED ON: OPENML-FLOW ยท F1 TELEMETRY ยท OPENCODE
      ๐Ÿš€ 05 // REPRODUCIBLE OPS
      ้›ฒ

      MLOps & Cloud Deployment

      Containerized sandboxing, deterministic artifact pipelines, automated CI/CD testing, and Linux telemetry.

      Docker Micro-VMs DVC MLflow Git & GitHub Actions Linux CI/CD
      VERIFIED ON: OPEN-MLPIPE ยท DOCKER CONTAINERS ยท CI WORKFLOWS
      โš”๏ธ 06 // SPATIAL & SHADERS
      ้ก

      Frontend Engineering & Interactive 3D

      Procedural WebGL shaders, Three.js spatial visualizers, GPU-accelerated canvas graphs, and luxury responsive UX.

      Three.js WebGL TypeScript JavaScript (ESNext) Canvas API GSAP Responsive Design
      VERIFIED ON: ATELIEROS ยท SAMURAI PORTFOLIO ยท OPENML FLOW
      EXPERIENCE & ENGINEERING MILESTONES // PROVEN RECORD

      Engineering Trajectory

      High-impact engineering milestones spanning founder-led architecture, upstream open-source contributions, and high-frequency real-time systems.

      1. 2025 โ€“ PRESENT
        FOUNDER & LEAD ARCHITECT OPEN SOURCE ยท MIT

        Founder & Lead Developer โ€” OpenML Initiative

        Architected OpenML Flow local-first visual IDE with real-time WebSocket telemetry. Designed Open-MLPipe deterministic DAG engine with SHA-256 state hashing contracts to eliminate data drift and guarantee zero-compute redundancy.

        Python 3.12 FastAPI WebSockets SHA-256 DVC Docker
        Explore OpenML Ecosystem โ†—
      2. 2025 โ€“ PRESENT
        LEAD ARCHITECT SPATIAL OS ยท MIT

        Lead Architect โ€” AtelierOS & Brain Nexus

        Engineered 3D spatial visual OS for autonomous AI agent swarms in sandboxed micro-VMs. Built ForceAtlas2 WebGL graph visualizer rendering 76K+ nodes from 67 merged repos with real-time multi-agent spatial coordination.

        Three.js WebGL LangGraph Multi-Agent Swarms Micro-VMs
        View Spatial OS Architecture โ†—
      3. 2024 โ€“ 2025
        ๐Ÿ† NTRO SIH FINALIST COMPETITIVE ALGORITHMS

        Research & Competitive Engineering โ€” SIH & LeetCode

        Built ThermoWatch satellite thermal anomaly clustering for NTRO Smart India Hackathon Challenge. Authored HARD-LEET-CODE solutions conquering complex dynamic programming, topological sorting, and graph tree traversals.

        PyTorch GeoPandas GDAL Dynamic Programming Graph Trees
        Inspect SIH Repository โ†—
      VERIFIED CREDENTIALS ยท 2026

      B.Tech Artificial Intelligence & Machine Learning (2024โ€“2028) ยท 5 Verified Professional Certifications ยท Upstream Open-Source Contributor

      ๐Ÿ“ฅ Download PDF